thesis

case study on artificial intelligence in agriculture

Share on FacebookShare on Twitter

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

• View all journals
• Explore content
• About the journal
• Publish with us
• Sign up for alerts
• Perspective
• Published: 23 February 2022

Responsible artificial intelligence in agriculture requires systemic understanding of risks and externalities

• Asaf Tzachor   ORCID: orcid.org/0000-0002-4032-4996 1 , 2 ,
• Medha Devare   ORCID: orcid.org/0000-0003-0041-4812 3 , 4 ,
• Brian King   ORCID: orcid.org/0000-0002-7056-9214 3 ,
• Shahar Avin   ORCID: orcid.org/0000-0001-7859-1507 1 &
• Seán Ó hÉigeartaigh   ORCID: orcid.org/0000-0002-2846-1576 1 , 5  

Nature Machine Intelligence volume  4 ,  pages 104–109 ( 2022 ) Cite this article

6852 Accesses

48 Citations

313 Altmetric

Metrics details

• Agriculture
• Science, technology and society

Global agriculture is poised to benefit from the rapid advance and diffusion of artificial intelligence (AI) technologies. AI in agriculture could improve crop management and agricultural productivity through plant phenotyping, rapid diagnosis of plant disease, efficient application of agrochemicals and assistance for growers with location-relevant agronomic advice. However, the ramifications of machine learning (ML) models, expert systems and autonomous machines for farms, farmers and food security are poorly understood and under-appreciated. Here, we consider systemic risk factors of AI in agriculture. Namely, we review risks relating to interoperability, reliability and relevance of agricultural data, unintended socio-ecological consequences resulting from ML models optimized for yields, and safety and security concerns associated with deployment of ML platforms at scale. As a response, we suggest risk-mitigation measures, including inviting rural anthropologists and applied ecologists into the technology design process, applying frameworks for responsible and human-centred innovation, setting data cooperatives for improved data transparency and ownership rights, and initial deployment of agricultural AI in digital sandboxes.

This is a preview of subscription content, access via your institution

Access options

Access Nature and 54 other Nature Portfolio journals

Get Nature+, our best-value online-access subscription

24,99 € / 30 days

cancel any time

Subscribe to this journal

Receive 12 digital issues and online access to articles

111,21 € per year

only 9,27 € per issue

Buy this article

• Purchase on SpringerLink
• Instant access to full article PDF

Prices may be subject to local taxes which are calculated during checkout

Similar content being viewed by others

Digital Regenerative Agriculture

The global divide in data-driven farming

Leveraging edge artificial intelligence for sustainable agriculture

The State of Food Security and Nutrition in the World 2020: Transforming Food Systems for Affordable Healthy Diets (FAO, IFAD, UNICEF, WFP, WHO, 2020).

Cole, M. B., Augustin, M. A., Robertson, M. J. & Manners, J. M. The science of food security. NPJ Sci. Food 2 , 14 (2018).

Article   Google Scholar  

Matson, P. A., Parton, W. J., Power, A. G. & Swift, M. J. Agricultural intensification and ecosystem properties. Science 277 , 504–509 (1997).

Quinton, J. N., Govers, G., Van Oost, K. & Bardgett, R. D. The impact of agricultural soil erosion on biogeochemical cycling. Nat. Geosci. 3 , 311–314 (2010).

Singh, R. B. Environmental consequences of agricultural development: a case study from the Green Revolution state of Haryana, India. Agricult. Ecosyst. Environment 82 , 97–103 (2000).

The State of the World’s Plant Genetic Resources for Food and Agriculture (FAO, 2010).

Semchuk, K. M., Love, E. J. & Lee, R. G. Parkinson’s disease and exposure to agricultural work and pesticide chemicals. Neurology 42 , 1328–1328 (1992).

Campbell, J. In Topical Research Digest: Human Rights and Contemporary Slavery 131–141 (Univ. Denver, 2008).

Nguyen, H. Q. & Warr, P. Land consolidation as technical change: economic impacts in rural Vietnam. World Dev. 127 , 104750 (2020).

Nilsson, P. The role of land use consolidation in improving crop yields among farm households in Rwanda. J. Dev. Stud. 55 , 1726–1740 (2019).

Du, X., Zhang, X. & Jin, X. Assessing the effectiveness of land consolidation for improving agricultural productivity in China. Land Use Policy 70 , 360–367 (2018).

Schmitz, A., & Moss, C. B. Mechanized agriculture: Machine adoption, farm size, and labor displacement. AgBioForum 18 , 278–296 (2015).

Wilde, P. Food Policy in the United States: An Introduction (Routledge, 2013).

Tadele, Z. Orphan crops: their importance and the urgency of improvement. Planta 250 , 677–694 (2019).

Lugo-Morin, D. Indigenous communities and their food systems: a contribution to the current debate. J. Ethn. Food 7 , 6 (2020).

Akinola, R., Pereira, L. M., Mabhaudhi, T., de Bruin, F. M. & Rusch, L. A review of indigenous food crops in Africa and the implications for more sustainable and healthy food systems. Sustainability 12 , 3493 (2020).

Jose, S. Agroforestry for ecosystem services and environmental benefits: an overview. Agrofor. Syst. 76 , 1–10 (2009).

Talaviya, T., Shah, D., Patel, N., Yagnik, H. & Shah, M. Implementation of artificial intelligence in agriculture for optimisation of irrigation and application of pesticides and herbicides. Artif. Intell. Agricult. 4 , 58–73 (2020).

Google Scholar  

Palacios-Lopez, A., Christiaensen, L., & Kilic, T. How much of the labor in African agriculture is provided by women? Food Policy 67, 52–63 (2017).

Alkon, A. H., & Agyeman, J. (eds) Cultivating Food Justice: Race, Class, and Sustainability (MIT Press, 2011).

Edmonds, E. V. & Pavcnik, N. The effect of trade liberalization on child labor. J. Int. Econ. 65 , 401–419 (2005).

Child Labour in Agriculture (International Labor Organization, 2021).

Lowder, S. K., Skoet, J. & Raney, T. The number, size, and distribution of farms, smallholder farms, and family farms worldwide. World Dev. 87 , 16–29 (2016).

Mehrabi, Z. et al. The global divide in data-driven farming. Nat. Sustain. 4 , 154–160 (2021).

Hennessy, T., Läpple, D. & Moran, B. The digital divide in farming: A problem of access or engagement? Appl. Econ. Persp. Policy 38 , 474–491 (2016).

Klerkx, L., Jakku, E., Labarthe, P. A review of social science on digital agriculture, smart farming and agriculture 4.0: New contributions and a future research agenda. NJAS Wageningen J. Life Sci. 90–91 , 100315 (2019).

Wolfert, S., Ge, L., Verdouw, C. & Bogaardt, M. J. Big data in smart farming–a review. Agric. Syst. 153 , 69–80 (2017).

Levins, R. & Cochrane, W. The treadmill revisited. Land Econ. 72 , 550–553 (1996).

Sontowski, S. et al. Cyber attacks on smart farming infrastructure. In 2020 IEEE 6th Int. Conf. on Collaboration and Internet Computing (CIC) 135–143 (IEEE, 2020).

Cyber-attack hits JBS meat works in Australia, North America. Reuters https://www.reuters.com/technology/cyber-attack-hits-jbs-meat-works-australia-north-america-2021-05-31/ (1 June 2021)

Sharma, A. $5.9 million ransomware attack on farming co-op may cause food shortage. Ars Technica https://arstechnica.com/information-technology/2021/09/5-9-million-ransomware-attack-on-farming-co-op-may-cause-food-shortage/ (21 September 2021)

Rahwan, I. et al. Machine behaviour. Nature 568 , 477–486 (2019).

Johnson, N. et al. Abrupt rise of new machine ecology beyond human response time. Sci. Rep. 3 , 2627 (2013).

Gold, E. R. The fall of the innovation empire and its possible rise through open science. Res. Policy 50 , 104226 (2021).

Majumdar, J., Naraseeyappa, S. & Ankalaki, S. Analysis of agriculture data using data mining techniques: application of big data. J. Big Data 4 , 20 (2017).

CGIAR GARDIAN Data Ecosystem https://gardian.bigdata.cgiar.org (CGIAR Platform for Big Data in Agriculture, 2021).

Yara and IBM. IBM https://www.ibm.com/services/client-stories/yara (accessed 18 August 2021).

Stilgoe, J., Owen, R. & Macnaghten, P. in The Ethics of Nanotechnology, Geoengineering and Clean Energy 347–359 (Routledge, 2020).

Theodorou, A. & Dignum, V. Towards ethical and socio-legal governance in AI. Nat. Mach. Intell. 2 , 10–12 (2020).

Kamle, S. & Ali, S. Genetically modified crops: detection strategies and biosafety issues. Gene 522 , 123–132 (2013).

Download references

Acknowledgements

This paper was made possible through the support of a grant from Templeton World Charity Foundation. The opinions expressed in this publication are those of the author(s) and do not necessarily reflect the views of Templeton World Charity Foundation.

Author information

Authors and affiliations.

Centre for the Study of Existential Risk, University of Cambridge, Cambridge, UK

Asaf Tzachor, Shahar Avin & Seán Ó hÉigeartaigh

School of Sustainability, Reichman University (IDC Herzliya), Herzliya, Israel

Asaf Tzachor

Platform for Big Data in Agriculture, CGIAR, Cali, Colombia

Medha Devare & Brian King

International Institute for Tropical Agriculture, CGIAR, Ibadan, Nigeria

Medha Devare

Leverhulme Centre for the Future of Intelligence, University of Cambridge, Cambridge, UK

Seán Ó hÉigeartaigh

You can also search for this author in PubMed   Google Scholar

Contributions

A.T., M.D., B.K., S.A. and S.Ó.H, developed the paper jointly and all contributed equally to the writing of the text.

Corresponding authors

Correspondence to Asaf Tzachor or Seán Ó hÉigeartaigh .

Ethics declarations

Competing interests.

The authors declare no competing interests.

Peer review

Peer review information.

Nature Machine Intelligence thanks Matthew McCabe and John Quinn for their contribution to the peer review of this work.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Reprints and permissions

About this article

Cite this article.

Tzachor, A., Devare, M., King, B. et al. Responsible artificial intelligence in agriculture requires systemic understanding of risks and externalities. Nat Mach Intell 4 , 104–109 (2022). https://doi.org/10.1038/s42256-022-00440-4

Download citation

Received : 20 August 2021

Accepted : 02 January 2022

Published : 23 February 2022

Issue Date : February 2022

DOI : https://doi.org/10.1038/s42256-022-00440-4

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

This article is cited by

Multi-disciplinary strategy to optimize irrigation efficiency in irrigated agriculture.

• Ayman Batisha

Scientific Reports (2024)

Rethinking digitalization and climate: don’t predict, mitigate

• Daria Gritsenko
• Bent Flyvbjerg

npj Climate Action (2024)

• Moussa El Jarroudi
• Louis Kouadio
• Said Hamdioui

Nature Sustainability (2024)

Who is responsible for ‘responsible AI’?: Navigating challenges to build trust in AI agriculture and food system technology

• Carrie S. Alexander
• Mark Yarborough
• Aaron Smith

Precision Agriculture (2024)

Large language models and agricultural extension services

Nature Food (2023)

Quick links

• Explore articles by subject
• Guide to authors
• Editorial policies

Sign up for the Nature Briefing: Anthropocene newsletter — what matters in anthropocene research, free to your inbox weekly.

Blog Artificial Intelligence: How could it transform agriculture?

• Artemis Project
• artificial intelligence
• Digital inclusion
• precision agriculture

AI is a controversial tool with powerful applications: including for how we grow our food. Below, we share examples of how researchers are already applying AI, and the considerations to sustainable and equitable use of this technology.

For most of us this year, the use of artificial intelligence (AI) has leapt from the realm of the possible to the inevitable. Sectors from healthcare to media are grappling with AI’s abilities, and agriculture is no exception. CGIAR scientists have already spent years exploring how AI can further efforts to sustainably grow resilient crops and ensure that farmers adapt to climate change.  

Three ongoing Alliance projects that draw on AI include: Tumaini (a smartphone app allowing banana farmers to solve 90% of major diseases and pests); Melisa (a chatbot that estimates the maize and wheat yield of Colombian farmers based on long-term weather predictions, soil and crop varieties, and sowing dates); and Artemis (computer vision technology systems that enable crop breeders to develop locally-adapted, climate-resilient varieties). 

Researchers in Tanzania are combining smartphone photos and AI to observe bean varieties in the field. Photo: David Guerena

Precision Agriculture: A Pathway to More Efficient Farming

Steve Mutuvi is a data scientist with a background in Large Language Models (the AI family behind ChatGPT ). With his work at the Alliance in Tanzania, he sees many opportunities to integrate these concepts into plant breeding:

“Applications of AI include yield optimization by predicting optimal planting, irrigation and harvesting times for maximizing yield while minimizing resource use and adapting to a changing climate. AI models can simulate how climate change will impact agriculture, and recommend strategies to make farming more resilient with a lower environmental impact.” 

Precision agriculture is a series of techniques using high-technology sensors and analytics to provide farmers with data to make informed decisions that minimize resource use and maximize yields. Precision agriculture tools are based on ‘ machine learning ’, which uses historical data to develop correlations between weather patterns, soil types, crop varieties, and external inputs, thus generating recommendations which can be provided to farmers in a variety of user-friendly formats.

One of the most valuable forms of AI-driven precision agriculture is crop monitoring : tracking crop health to offer tailored guidance for farmers to ensure a good harvest, simultaneously safeguarding their livelihoods and reducing waste.

Disease Diagnosis and Mitigation: Tumaini

Our first case study is ‘ Tumaini ’, which in Swahili means ‘hope’. Developed by Alliance scientists and local banana farmers, this easy-to-use app applies machine learning (analysis of phone, drone, and satellite images ) to detect early signs of five common diseases and one major pest , which in the past has caused many farmers to lose their season’s entire harvest. It is simple to use: farmers upload a photo of an affected crop, the app compares the photo with a database of images categorized by location, and finally the app offers the farmer a diagnosis and a series of recommendations to solve the problem. Already in 2019, Tumaini trials in Colombia, the Democratic Republic of the Congo, India, Benin, China and Uganda returned a 90% rate of successful disease and pest detection . According to project leader Michael Selvaraj : “Tumaini AI has transformed GPS data into a vivid portrait of banana plant health worldwide: from 6,000 in 2013 to over 18,000 entries today”. Its latest step is to include another important crop: beans. 

Artificial intelligence helps banana growers protect the world’s most favorite fruit

Melisa chatbot - the 'oracle' for Colombian farmers

The oracle of yield forecasting: melisa.

Using a different communication method to support maize and wheat producers in Colombia, in 2022 the Alliance launched ‘ Melisa ’: an AI-driven chatbot that provides accurate agro-climatic forecasts , allowing farmers to plan harvests, and prepare their products for market. Melisa is available to farmers on WhatsApp, Facebook and Telegram , and in the form of a chat, farmers can ask ‘Melisa’ for both short- and long-term weather predictions, as well as estimations of their maize and wheat yield for the coming season. Melisa uses a machine learning system that predicts farmers’ outputs by analyzing their soil type, the seed varieties used, the days they sowed the seeds, as well as the past and predicted climatic conditions. This information allows farmers to plan optimal harvesting dates, and estimate their income for the season.

Farmer-driven Phenotyping: Artemis

AI’s information-processing capacities can also empower farmers to cultivate more locally-adapted crops through faster and more accurate phenotyping (observing how crops perform to select the most promising varieties), accelerating the traditional process of plant breeding which farmers have done manually for thousands of years. AI speeds up this selection process by analyzing and comparing thousands of images of crop varieties, as they grow in the field. The Artemis project - a collaboration between the Alliance and the Alphabet company Mineral - has been using rovers and smartphones  to help farmers and crop breeders identify the most productive and resilient seed varieties based on their location and unique growing conditions, at sites from Colombia to Tanzania. 

Meet Don Roverto and Tatiana, the robotic vehicles helping to breed better beans

From Silicon Valley to Tanzania: Putting AI to work for smallholder farmers

Further areas for exploration.

As well as crop monitoring and advisory services, another promising example of AI in agriculture is the installation of underground soil humidity sensors that could allow farmers to estimate irrigation needs , thus helping them use resources efficiently. AI could also help design automated smart irrigation systems – a complex process yet to be upscaled. Furthermore, advanced robotic technologies could automate manual tasks such as sowing and harvesting , thus reducing the time-consuming labor needs of farming, potentially saving money for farmers and reducing human error.

The Complexities of AI in Agriculture

Given that artificial intelligence is rapidly evolving and still yet to be fully understood, there are potential challenges for its implementation, including: 

• Data privacy and security: AI relies on collecting large amounts of data on farmers’ practices and growing conditions, which could make them vulnerable to identity theft or the disclosure of confidential information. Data sharing risks could also make farmers wary of adopting AI technologies, making it hard to upscale the benefits these tools may provide.
• Dependence on technology: If farmers come to rely on AI tools for their planning, any disruption or technical failures could put their operations and productivity at risk.
• Unequal access: Depending on pricing, some small-scale farmers may not have the financial capacity to afford access to AI technologies. This could cause inequality in favor of larger producers, putting small-scale farmers at a disadvantage, leading to further inequality; this raises the question of how such technologies could be made available to these farmers free of charge.
• Job displacement: The potential of automating farming labor by using rovers or robots could put farmworkers at risk of losing their jobs, potentially creating wider economic and social difficulties.
• Monoculture vs agrobiodiverse systems: AI’s focus on minimum input and maximum efficiency and yield may not work for farmers who intend to adopt agroecological practices. Creating biodiverse farming systems with approaches such as agroecology is seen as an important way to restore and protect ecosystems that have been affected by industrial agriculture and chemical inputs. A variety of farming methods must be explored, but considering the importance of agrobiodiversity, AI tools could create a bias in favor of monoculture.
• Ethical concerns: AI’s potential to contribute to genetic modification provokes many ethical questions and debates. While the possibility of modifying seeds to increase yields may contribute to food security – one of the world’s most pressing concerns – even natural seed selection processes through phenotyping may reduce public trust and acceptance of these technologies.
• Environmental impacts: While AI could increase resource efficiency by predicting accurate needs for irrigation and other inputs, if adopting these tools creates a bias towards monoculture and incentivizes chemical inputs, this could reverse the gradual advances of agrobiodiversity. Furthermore, upscaling these tools also requires resources needs, possibly leading to yet-unknown negative environmental impacts.

How are Alliance's researchers addressing these risks? Berta Ortiz is a specialist in human-centered design who is integrating participatory approaches into Artemis’ work in Tanzania. She says:

“Inclusive design can overcome some of the issues faced by artificial intelligence by identifying the diverse needs of those that will use the AI tools, ensuring that they suit their unique contexts.”

The first step to recognizing and avoiding problematic applications of AI is to work with end users - local farmers and breeders - from the very beginning . 

Overall, while artificial intelligence has massive potential to improve food production, it is simply an additional tool for researchers, breeders, and farmers, and requires careful use to yield the best results. As the importance of AI grows in all sectors, CGIAR and the Alliance will continue to develop and deploy these tools for the benefit of farmers first, as part of the transition towards sustainable and equitable food systems. 

More about these AI Tools

Tumaini app

Melisa chatbot

Artificial intelligence and new business models in agriculture: a structured literature review and future research agenda

British Food Journal

ISSN : 0007-070X

Article publication date: 12 July 2023

Issue publication date: 18 December 2023

Artificial Intelligence (AI) is a growing technology impacting several business fields. The agricultural sector is facing several challenges, which may be supported by the use of such a new advanced technology. The aim of the paper is to map the state-of-the-art of AI applications in agriculture, their advantages, barriers, implications and the ability to lead to new business models, depicting a future research agenda.

Design/methodology/approach

A structured literature review has been conducted, and 37 contributions have been analyzed and coded using a detailed research framework.

Findings underline the multiple uses and advantages of AI in agriculture and the potential impacts for farmers and entrepreneurs, even from a sustainability perspective. Several applications and algorithms are being developed and tested, but many barriers arise, starting from the lack of understanding by farmers and the need for global investments. A collaboration between scholars and practitioners is advocated to share best practices and lead to practical solutions and policies. The promising topic of new business models is still under-investigated and deserves more attention from scholars and practitioners.

Originality/value

The paper reports the state-of-the-art of AI in agriculture and its impact on the development of new business models. Several new research avenues have been identified.

• Artificial intelligence
• Agriculture
• Business models
• Literature review
• AI-Based applications
• Sustainability
• Sustainable business models
• Agricultural policies

Cavazza, A. , Dal Mas, F. , Paoloni, P. and Manzo, M. (2023), "Artificial intelligence and new business models in agriculture: a structured literature review and future research agenda", British Food Journal , Vol. 125 No. 13, pp. 436-461. https://doi.org/10.1108/BFJ-02-2023-0132

Emerald Publishing Limited

Copyright © 2023, Alberto Cavazza, Francesca Dal Mas, Paola Paoloni and Martina Manzo

Published by Emerald Publishing Limited. This article is published under the Creative Commons Attribution (CC BY 4.0) licence. Anyone may reproduce, distribute, translate and create derivative works of this article (for both commercial and non commercial purposes), subject to full attribution to the original publication and authors. The full terms of this licence may be seen at http://creativecommons.org/licences/by/4.0/legalcode

1. Introduction

Artificial intelligence (AI) is a growing technology that is attracting the interest of both academics and practitioners ( Arora et al. , 2022 ). Several definitions of AI have been given periodically, redefining the concept according to the latest advancements. In one of the earliest definitions, Kok et al. (2002 , p. 2) called it “an area of study in the field of computer science concerned with the development of computers able to engage in human-like thought processes such as learning, reasoning, and self-correction.”

Today, AI is widely employed in several fields, and its applications are progressing, becoming more precise and performant, including manufacturing ( Bagnoli et al. , 2022 ), healthcare ( Cobianchi et al. , 2023 ; Loftus et al. , 2020 ), banking and finance ( Doumpos et al. , 2023 ), aviation ( Kulida and Lebedev, 2020 ) and hospitality ( Goel et al. , 2022 ). Among its several applications, AI is being employed in the agricultural field as well, with the aim of improving yield, efficiency and profitability ( Dal Mas et al. , 2023 ) and developing economic forecasts ( Chu et al. , 2019 ; Lebelo et al. , 2022 ). AI in the agricultural sector includes innovative technologies such as field sensors, drones, farm management software tools, automated machinery and water and fertilizer management solutions ( Arora et al. , 2022 ; Misra et al. , 2022 ; Romanello and Veglio, 2022 ; Trivelli et al. , 2019 ). In this category, new innovative farming techniques such as vertical farming ( Biancone et al. , 2022 ; Musa and Basir, 2021 ; Saad et al. , 2021 ), aquaculture, insect breeding and precision agriculture can be included ( Dal Mas et al. , 2023 ; Trivelli et al. , 2019 ).

AI in agriculture can play a strategic role. Indeed, at a global level, the agricultural sector has a value of 3,6 trillion dollars, providing the 4% of the global gross domestic product (GDP) with a stable measure during the last twenty years. Moreover, in some developing countries, it accounts for more than 25% of GDP ( FAO, 2022 ). Such a critical industry stands as a food and energy base of the new economy, mainly because it ensures food security ( Magasumovna et al. , 2017 ).

Still, various implicit problems have been historically challenging the agricultural sector. The first of such issues is undoubtedly the number of workers which is significantly collapsed with a progressive difficult-to-employ workforce. For instance, between 2000 and 2022, the global workforce employed in agriculture collapsed from 40% to 27%, representing a reduction of 177 million people ( FAO, 2022 ). These data underline the technological impact in this field in the last century, with a food production increment per person less than proportional with the population growth; this previous more than doubled between 1950 and 1998 ( Sunding and Zilbermanof, 2001 ). In the last years, there has been a similar trend with an increasing population but decreasing productivity caused by climate change and desertification, with a decline of 134 million hectares of cultivated land between 2000 and 2020 ( FAO, 2022 ). For these reasons, achieving food security in a sustainable way is one of the objectives included in the United Nations (UN) 2030 Sustainable Goals with the Zero-Hunger program ( European Commission, 2017 ). A country can be considered food secure “if food is available, accessible, nutritious and stable across the other three dimensions” ( Musa and Basir, 2021 , p. 3087). According to the latest FAO World Food and Agriculture – Statistical Yearbook (2022) , in 2021, 770 million people were undernourished, with an increment of 150 million from 2020 ( Wijerathna-Yapa and Pathirana, 2022 ). As a result, it emerges a growing need to modify agricultural methods and available technologies so that “maximum crops can be attained and human effort can be reduced” ( Saad et al. , 2021 ).

Innovation technology, digitalization and AI could, therefore, represent some of the ways and strategies to mitigate the abovementioned issues, achieve sustainability goals and manage the climate change challenge ( DiVaio et al. , 2020 ; Yela Aránega et al. , 2022 ). For this reason, the topic of AI applications in agriculture is worth investigating as an opportunity to address some of the cited problems creating new business scenarios in the agricultural sector ( Amoussohoui et al. , 2022 ). While the digital revolution has already changed the world ( Bresciani et al. , 2018 , 2021b ), only in the last years the agricultural sector has started to integrate information and communication technologies in traditional farming with the aim of improving crop yield efficiency, reducing costs and optimizing process inputs with the usage of data ( Boursianis et al. , 2022 ).

AI has proved its capability to lead to new business models ( Dal Mas et al. , 2021 ; Wamba-Taguimdje et al. , 2020 ). A business model can be defined as “a modeling and representation tool [which] represents a dynamic system, made of elements coherently in the relationship between them. The business model is used to understand the logic of an organization for the value creation” ( Bagnoli et al. , 2018 , p. 56). Creating new business models in agriculture could support the sector's development, providing solutions to the abovementioned issues, also under a sustainability lens ( Biancone et al. , 2022 ; Shukla and Sengupta, 2021 ).

What can be the contribution of AI to the agricultural sector, especially in the creation of new business models?

What research implications emerge?

The paper is organized as follows. Section 2 reports the methodological remarks in conducting the study. Section 3 summarizes the main findings of the literature analysis. Section 4 discusses the main results of the research questions in a critical way. Section 5 depicts the limitations and future policy avenues.

2. Methodology

2.1 selection criteria.

The paper adopts a structured literature review (SLR) defined by Massaro et al. (2016 , p. 767) as “a method for studying a corpus of scholarly literature, to develop insights, critical reflections, future research paths, and research questions.” As recommended by the methodological articles by Massaro et al . (2016) and Kraus et al . (2020 , 2022) , the authors prepared a literature review protocol to guide the analysis creating a framework to select, analyze and assess the academic production to ensure the study “to be reproducible, well-evidenced, and transparent, resulting in a sample inclusive of all relevant and appropriate studies” ( Kraus et al. , 2022 , p. 2579).

In accordance with previous studies ( Secinaro and Calandra, 2021 ), the scientific database Scopus was employed to find relevant contributions to be analyzed. The search key “Artificial intelligence AND Agriculture AND Business model” in the title, abstract or keywords, conducted on September, 13th 2022, led to 73 total contributions [1] . As recommended by previous articles ( Bresciani et al. , 2021a ), to cross-validate the results, the same search query was verified in the EBSCO Business Premier and Web of Science (WoS) datasets, leading to the same results.

As the initial number of documents was not too extensive, the authors decided to keep all the source types to be assessed in more detail by reading the provided abstracts to ensure eligibility. Interestingly enough, several published conference proceedings appeared in the document list. Most literature reviews tend to exclude such sources, as they are considered less rigorous than articles published in peer-reviewed journals. Still, when considering cutting-edge research topics like the ones connected to the development of modern technologies, early results may be shared at conferences before being sent out for a more rigorous peer review journey. Therefore, the authors decided to consider conference proceedings eligible in the sample as they provide “insights into the areas of debate that will later appear in academic journals” ( Dumay et al. , 2016 , p. 168).

After reading all the abstracts, of those 73 journal papers, conference proceedings, books, book chapters and editorials, 45 have been considered appropriate for the analysis, while 28 were considered off-topic, as they did not deal with the theme under a managerial or economic lens, rather an information technology or computer science one. Of these 45 eligible works, 6 of them were not retrieved, while the other 39 were coded using the Nvivo software. During the codification process, two additional papers were excluded because they were off-topic after eligibility. The final sample of 37 works was considered appropriate, as very close to the target of 40 articles which “indicates that the domain has reached sufficient maturity for review” ( Paul et al. , 2021 , p. 4).

The following Figure 1 reports the selection process following the PRISMA methodology ( Page et al. , 2021 , Schünemann et al. , 2021 ).

2.2 Coding framework

In coding the items using Nvivo, several nodes were gathered from previous studies, while others were decided following an extensive discussion among the authors, considering the specific field of investigation.

The first node refers to the type of authors dividing them among academics, practitioners and collaborations ( Dal Mas et al. , 2020 ). The second node refers to the source type. The third node maps the location where the study is conducted, grouping countries by continent ( Massaro et al. , 2015 ). The fourth group of nodes refers to the employed research method ( Paoloni et al. , 2021 ).The fifth node concerns the agricultural sector, while the sixth category lists the problems to solve and the objectives to reach. In this last node, the sub-nodes were added while coding the papers, employing an open coding approach. The seventh node analyzes the technology used and reported in the studies. The eighth node group maps the application in agriculture, while the ninth node focuses on identifying sources which treat a business model. The ninth node is about the eventual possibility of leading a new business model. The tenth node analyzes the eventual connection with sustainability issues. Last but not least, the last nodes refer to the presence of research, practice and policy implications.

Table 1 reports the bibliographic details of the 37 articles and conference proceedings which were included in the literature review. While the earliest work dates back to 2005, twenty-four contributions (65% of the total sample) were published after 2017, highlighting the increasing interest in this topic in the last few years.

The following Table 2 underlines the results of the Nvivo coding, following the defined framework.

Concerning the node about authorship, authors are mainly represented by academics with twenty-five contributions. Interestingly, eight works result from a collaboration between scholars and practitioners. Five articles are authored by practitioners, mainly belonging to institutional agricultural research centers.

Twenty-one sources are represented by journal articles, while sixteen are conference papers.

Concerning the location of the study, twenty-four sources specify the place where the investigation was conducted, while thirteen papers have no specific area as they refer to specific technological solutions or algorithms. Considering the documents that do declare the location of their investigation, eleven sources are focused on Asia and seven on America (including both North and South America). Six references refer to Europe, while Africa and Oceania have respectively two papers for each continent. However, there is not an absolute predominance. Therefore, it may be claimed that the sample is well representative worldwide.

When referring to the research methodology, the vast majority of the sources (26 papers, equal to 70% of the total sample) are represented by case studies, while the remaining eleven papers are literature reviews. Still, the formers are mainly represented by theoretical investigations which focus on a new technological application presentation and discussion. Neither success (or failure) stories nor business translation experiences are reported.

Focusing on the agricultural sector, fifteen sources relate to the cultivation of plants, while some argue about the business in general terms. Animal production is treated in six papers, while only one article discusses fish farming. All in all, there seems to be good coverage of topics, which expresses the various interests both from general and specific research groups.

Regarding the specific issues and problems that stimulated the analysis, the goal of a significant number of sources refers to increasing efficiency and maximizing the farm return, with twenty-six papers. The need to manage the environmental impact and the external changes are treated in twenty-four articles. Moreover, nineteen papers discuss the issue of predicting and managing farm complexity, but, at the same time, great relevance is given to the food-security problem, discussed in nine sources. The research piece by Ahmed et al. (2022) is an example of this last issue. In the paper, the authors predict that climate change, especially global warming and increasing temperatures, could put half of the global population in trouble due to the declined crop productivity. Only two articles report other objectives. The different types of issues are strictly connected, with some articles arguing about more problems together. As an example, managing farm's complexity may lead to an increase in efficiency and profitability, creating a sort of turbo effect. For instance, Bogomolov et al. (2021) highlight the connection between the need to improve yields with the desertification problem and the related reduction of pesticides. The following Table 3 describes in more detail each sub-node with more specific problems to be taken into consideration.

Concerning the technologies that are mentioned within the papers, a significant number of sources treat Decision Support Systems (DSS), which stands as the most present technology. Only nineteen articles specifically refer to AI and Machine Learning. Other technologies with great relevance that are reported in the articles are represented by Big Data Analytics and the internet of Things (IoT). Other less-discussed technologies are represented by drones and robots (eight papers), cloud computing (seven articles), geographical indication systems and other technologies (six papers). Finally, biotechnology, Blockchain and autonomous devices are named in three pieces. Although the research has been based on AI as the leading keyword, the selected articles report several kinds of technologies, given their outstanding level of integration and complementarity. DSS is the most used technology because it represents the predecessor of AI. Within AI, we find all the sources which discuss Machine Learning and all its specializations, such as Artificial Neural Networks and Deep Learning.

The node about the applications in agriculture allowed the investigation of the proposed applications in the agriculture field, leading to four main results. The first and the most treated is precision farming and other types of agronomic applications discussed in twenty-four papers. Agronomic planning and economic applications are reported by twenty-one sources. Less common applications are represented by water optimization with environmental management and supply chain applications with traceability systems, which are discussed respectively in fifteen and five papers. The following Figure 2 reports the main AI-based applications, dividing them into categories and naming those which were cited by more than two articles.

There seems to be a link between the applications and the problems to solve; the former tries to find feasible solutions by employing innovative and practical ways. For instance, Li  et al. (2022) propose an Artificial Internet of Things (AIOT), which permits to obtain crop growth parameters in real-time, supporting farmers in managing farm complexity and unpredictability. Furthermore, the proposed solution makes intelligent recommendations for fertilization, crop disease detection and irrigation optimization. Another example is represented by Skobelev et al . (2019) , who offer several precision farming solutions with the objective of increasing productivity and efficiency of crop production. Moreover, benefits include cost reductions along the chain of production. The following Figure 3 shows the link between the problems to be solved and the applications, underlining several connections.

One of the critical points of the analysis was to understand the type of business models reported by the articles as a consequence of the application of AI. Interestingly enough, despite mentioning the words “Business model” either in the title, abstract and/or keywords, most sources do not report any kind of business model. Indeed, only seventeen papers responded positively to this question. Among such sources, the most discussed business model is surely represented by smart farming with thirteen articles, followed by data-driven business models with eight papers and, finally, the general industry 4.0 business model with only two sources. However, findings are very connected to each other because both data-driven and smart farming are part of the more inclusive industry 4.0 business models, which permit enhancing the value proposition, solving critical factors and delivering meaningful experiences to customers ( Bagnoli et al. , 2022 ; Pietrewicz, 2019 ).

The following node is connected to the previous one, investigating the possibility of AI leading to a new business model. Again, most articles do not mention any type of new business model, with only six papers trying to address such a challenge. Among these articles, two sources propose a platform business model used for the food supply chain where the key participants of the agriculture industry can sell and offer their products and services with the use of smart contracts. Moreover, they can exchange data by enriching a common dataset ( Skobelev et al. , 2019 ; Sood et al. , 2022 ). The same number of sources propose an Agritech 4.0 business model with an integrated food supply chain, where the new technologies permit to integrate both food production and food distribution, ensuring transparency, traceability and customer satisfaction ( Eashwar and Chawla, 2021 ; Wolfert et al. , 2017 ). Finally, supply chain management 5.0 and new information-based systems based on traceability are reported. The former proposes a new supply chain solution based on driverless autonomous vehicles for transporting and smart contracts with face recognition, while the second treat a new system based on recommended guidelines and documentation requirements for decision-making processes to ensure traceability along the chain ( Ahamed and Vignesh, 2022 ; Li et al. , 2017a ). However, an interesting consideration is that all four new solutions are inherent to the food supply chain and to the need to reduce complexity through technology integration. These efforts are also addressed to reduce global food waste along the food chain, which, according to a 2011 FAO report, equals one-third of the global production ( UN Environment Programme, 2021 ).

Another point of analysis referred to a potential connection with sustainability issues. Interestingly, most articles discuss sustainability issues, with only fourteen pieces not considering environmental or social topics. Five different kinds of sustainability issues can be reported. The first and the most treated is the use of fertilizers, nitrates and heavy metals, which pollute agricultural soil and water (eight references, equal to 35% of the total sample) and after the need to reduce the use and waste of water in the agricultural sector. The other topics are related to the need to produce climate-oriented and ecologically friendly applications, the need to achieve the food-security in a sustainable way and the need to make sustainable the production of some types of foods which actually heavily impact the environment.

Concerning the advantages gathered from the application of AI, almost all the sources (34 papers equal to 92% of the total sample) explain the benefits of the new technology implementations in the agricultural sectors. The most discussed advantages are represented by the organizational advantages and the decision-making support. Other advantages are related to the efficiency benefits and the productivity increase, while only two pieces for each pro speak about environmental benefits and food-safety issues with the possibility to control food compliance easily.

Another node concerns the disadvantages. Interestingly enough, just seven articles discuss the cons, with the majority of the sources not discussing such issues. Some examples are represented by the inevitable loss of income related to the compliance with water restrictions for small vineyards farms or the fact that some irrigation decision-making systems are crop specific for a given area with a consequent great complexity to generalize the methods for other crops and areas ( Carmona et al. , 2011 ; Nada et al. , 2014 ).

About the barriers that can limit the spreading of new technology, only fourteen papers discuss innovation barriers. The two most significant ones are the farmers' lack of technical knowledge about information and communication technologies (ICT) and emerging technologies and the limited equipment, Internet access, storage capacity and high-quality data, especially in developing countries. Bogomolov et al. (2021) , for instance, highlight the lack of qualified personnel and high-quality Internet access as two of the main problems in the field of applied digital technologies in the Russian agricultural industry, which hinder productivity and efficiency improvement. Six papers deal with the high investment cost and low perceived effectiveness. From such a perspective, Wakjira et al. (2021) analyze a case of precision beekeeping in Indonesia and Ethiopia, highlighting the impossibility of using commercial systems of remote bee colony monitoring because local beekeepers cannot afford them. Finally, some sources treat the mismatch between farmers' practical needs and the available applications, data control and data security problem, the lack of integration of the food supply chain, the large energy consumption of these innovations and the user psychological barriers to the implementation.

Concerning the research implications, only sixteen papers report any, ten concerning the need to extend and integrate the study with new data types or focus on new related issues. The remaining sources advocate testing the proposed method, analyzing profoundly new aspects and finally explaining the need to develop new solutions and technologies.

Concerning the practical implications, twenty-six sources lead to some practical consequences, especially for farmers. Such a topic appears to merge theoretical insights and practical applications, and it welcomes practical user solutions. Themes include the potential to help farmers in the decision-making process, support everyday farming operations, and to increase efficiency and effectiveness. No surprise AI is historically strictly connected to decision-making support, with a substantial increase in the last years as a consequence of the availability of new data sources and the decreasing cost of technological tools ( Secinaro et al. , 2022 ). AI is able to make the needed changes in the decision-making process supporting new ways to identify the critical variables of the decision space, the interpretation of the process, the final result and the several alternatives with the possibility to replicate the transaction, reducing time and costs ( Shrestha et al. , 2019 ). Another significant practical implication concerns the possibility of helping farmers manage the implicit farm unpredictability in the planning process. Finally, some sources provide farmers with new emerging and integrated technologies to develop and test.

Last but not least, only nine papers report some policy implications, mainly represented by government involvement. Four articles explain as governments should use agricultural data from fields to improve policy-making decisions, learning from data to create better future forecasts. At the same time, four sources recommend governments subscribe to new investment plans to enhance the technological transition, for instance, in publicly accessible digital infrastructures, protecting platform workers' rights and customer privacy ( Chiles et al. , 2021 ). Other contributions encourage policymakers to support farmers in technology knowledge acquisition by creating advisory units composed of experts ( Sood et al. , 2022 ), and to support social innovation by engaging the younger generations ( Wakjira et al. , 2021 ).

4. Discussion

As already explained in the introduction, this study aims to examine and better understand the role of AI in the agricultural sector, focusing on the possibility of AI creating new business models and understanding the research implications.

4.1 State-of-the-art and new applications of AI in the agricultural field

In addressing the first research question, results depict a lively situation characterized by a high speed of change and development. In such a perspective, findings report many collaborations and the presence of papers authored by practitioners, which looks unusual in academia, where the academic-practitioner divide exists in many fields ( Massaro et al. , 2018 ). Such a finding suggests that this topic represents an advanced and high-technical field where theory is strictly connected to practical applications. Innovation happens first in practice and can lead then to academic works and reasoning. Therefore, the practitioners' role in the field is extremely important. Academics are so invited to partner with managers and private companies to study the advancements and innovations in the field, share the best practices and business cases and suggest methodologies to assess the technology, measure and report its impacts, suggesting practical, research and policy implications.

Moreover, the unusually high number of conference proceedings extracted from Scopus and included in the sample can be connected with the previous point concerning the role of practitioners. Indeed, when high-technological fields are under the academic lens, scholars tend to present an early-stage draft of their works at conferences, getting feedback from their fellows before submitting their articles for peer review. In the case of AI, it seems like the implementation of new technologies and new agricultural innovations are initially presented during conferences and only after being discussed in the academic literature. Conferences, congresses and professional and institutional meetings then become relevant places where the latest advances are presented, shared and discussed.

Regarding the types of technology, although the research key used in Scopus specifically mentioned the words “Artificial Intelligence,” twelve different kinds of technologies are reported. This fact may be explained as AI is only a part of a greater system of Industry 4.0 digital paradigms used as methods to develop analysis and prediction with further disciplines such as data science, electronic engineering and so on. For this reason, AI is a technology that may be fully integrated with other digital paradigms such as smart manufacturing, autonomous and collaborative robots, augmented and virtual reality, industrial IoT, cloud computing, big data analytics and cybersecurity, permitting to reach economies of scale with high levels of personalization. A complementarity among technologies emerges. Notably, particularly significant seems the relationship between AI and IoT, merged by Li et al. (2022) in the new term “AIOT.”

As already reported in the results, a relevant number of practical implications are related to decision-making support provided by these new technological implementations. At this point, the farmers' capacity to use these innovations in the right way looks fundamental. About the practical application in agriculture, precision farming emerges as a new method to increase efficiency and reduce losses. Precision agriculture could be defined as a new method of smart agriculture which permits connecting resources with needs, growing, in this way, efficiency and productivity while also reducing the environmental impact and the unpredictability of the farm return ( Boursianis et al. , 2022 ).

4.2 Research methods

Also the research methods second the academic-practitioner alliance in this field. Indeed, the research methods adopted underline how case studies play a vital role in the literature. Most of these cases do not “tell” the success or failure stories of companies or farmers. Still, they assess and discuss new innovations and their practical applications, still with little emphasis on the consequences for the business, the technology acceptance and ethics dynamics and the need to engage in new educational paths to gain new competencies and skills. That is also why most cases do not refer to any specific geographical location, as new applications may be employed everywhere. Even if such a development may sound “natural” considering the field and the speed of change, the scientific community belonging to the management, organization and accounting fields should contribute to the multidisciplinary debate by sharing more success stories, even comparing multiple cases, highlighting the advantages and disadvantages of some solutions. In addition, another key issue may be represented by the rate of acceptance of these new applications in practice. Therefore, quantitative research methods like surveys and questionnaires should be tested by agricultural operators, who directly use the technological application during their everyday operations, or Delphi panels to assess the potential of some new solutions, even in their early stage of development. Researchers should target small and medium farmers, who represent the majority of agricultural enterprises in several continents, but who often have little capital to invest and a lower level of technological knowledge. The latter is indeed reported in the barriers as one of the most significant hurdles to digital transactions. For this reason, trade associations and agricultural consortia may organize open recurring conferences to diffuse and disseminate the opportunities brought forth by AI and Industry 4.0 to all the operators in this field.

4.3 Geographical areas

Another interesting result comes from the locations where the studies were conducted. The topic is widely diffused around the world, with a concentration in Asia, which is actually the hub of global innovation. Asian countries are implementing several policies to support innovation, start-ups and the creation of business incubators ( GT staff reporters, 2023 ). From the yet limited sample, Europe is actually even behind the USA and South America. Furthermore, while Africa appears in the sample with just a few contributions, it may represent an exciting outlet for technology providers, given its significant presence of arable land and the actual low level of technological advancement. While more barriers may be present than elsewhere (especially concerning the lack of infostructure and the financial investments needed), Africa stands as a continent whose development may largely benefit from AI.

4.4 Business model innovation

In addressing the business model topic, interesting thoughts should be made. Even the papers that somehow mention the matter do not clearly explain the business model name. Interestingly, there is a lack of business model definition in all these papers. Still, new technologies are supposed to be the triggers of new business models with technology-driven innovation ( Biancone et al. , 2022 ; Bresciani et al. , 2021a ; Secinaro et al. , 2022 ). The discouraging results open up exciting research avenues in mapping and defining new business models in the agricultural field, their unique features, the opportunities they may bring, the outcomes, the operational consequences and needs and the chance to involve different stakeholders with relevant implications for business practices as well. Researchers may borrow some sound results and experiences scouted in other fields.

4.5 Connection to sustainability issues

Although the research did not mean to focus on the sustainability issue in agriculture, findings show that the two topics are highly related. Farmers should take into consideration the environmental impact of their activity. Moreover, there is an influence of the environmental variables on the seasonal outcome, which determines the farm profit. This is intrinsically at the core of farm management, but now, with digital technology support, it is possible to manage farm unpredictably. A new innovative paradigm is given by vertical farming, a new way of production which permit to control all the agricultural variables using the so-called Controlled Environmental Agriculture together with the nature co-design, increasing resilience and circularity through hydroponic cultivation and advanced led lighting systems ( VanGerrewey et al. , 2022 ) AI can create new sustainable business models improving the technical-scientific quality of the production system. For this reason, a focus should be placed on applications which provide both profit and sustainability ( DiVaio et al. , 2020 ), also leading to new sustainable business models for value creation.

Starting from the analysis of the results, the following Table 4 summarizes the new research avenues for each of the identified macro-topics.

5. Conclusions

The article underlines the potential role of multiple AI solutions in disrupting the agricultural sector by offering sound opportunities to farmers and entrepreneurs in the field to support their decision-making process and increase the farm's profitability. Still, literature and practice are in progress, with more solutions and applications being developed and tested and more opportunities to disrupt business models, even fostering sustainability practices. Academic engagement with professionals stands as a relevant strategy to stimulate the debate, study the managerial and organizational dynamics and suggest and spread new business procedures.

Several new research avenues have, therefore, been suggested: from the employment of both quantitative and qualitative research methodologies to a deeper collaboration with practitioners, from spreading best practices and lessons learned to comparative studies among different contexts and countries. Promising research themes include the features of potential new business models, the degree of technology acceptance up to the educational needs of farmers and communities, among others.

5.1 Limitations

As with all studies, this has limitations. Even if the methodology can be considered rigorous and replicable, the sample of analyzed sources is small, with the use and cross-checking of a limited number of scientific datasets. In addition, the coding process may leave room for subjectivity. Moreover, the speed of technology development and the quantity of new academic pieces published every month may impact the validity of the results. Such limitations may lead to further research opportunities to frame the phenomenon and its fascinating yet helpful outcomes, also scouting the so-called “grey literature” coming from governmental institutions, consultancy firms, patent datasets, professional magazines and reviews and recognized practice blogs, as reported by other studies ( Dal Mas et al. , 2023 ; Secinaro et al. , 2022 ).

5.2 Policy implications

While practice implications are more connected to technological advances and the application of new business models, some relevant policy implications emerge.

Policies may be linked to the identified barriers in the practical applications of these new AI-based solutions. These barriers include the lack of farmers' ICT knowledge and technology acceptance dynamics. Findings explain how these barriers in some cases are agricultural specific, such as in the case of the complexity and lack of integration of the food supply chain, but the majority are represented by general barriers to the implementation, which are common to all other sectors. As already suggested, the agricultural field could borrow or adapt solutions created and already implemented for other sectors solving a significant number of problems. For this reason, policymakers should stimulate the collaboration between key agricultural stakeholders and actors involved in different sectors, to solve the general barriers to the implementation. Governments play a vital role in fostering the creation of new general solutions and the adaptation of existing systems, including the availability of dedicated funds or tax privileges to support farmers (especially small-sized companies) in technology acquisition. Knowledge translation and dissemination initiatives involving multiple stakeholders like agricultural consortia, technology providers, research institutes and universities could help to overcome the acceptance issues and the understanding of the new opportunities for the single farm and the more comprehensive ecosystem.

The process of article selection following the PRISMA methodology

The main AI-based applications in agriculture

The link between problems and applications

Bibliographic details of the included works

#	Authors	Title	Year	Source title	Ref
1	Ahmed M., Hayat R., Ahmad M., ul-Hassan M., Kheir A.M.S., ul-Hassan F., ur-Rehman M.H., Shaheen F.A., Raza M.A., Ahmad S.	Impact of Climate Change on Dryland Agricultural Systems: A Review of Current Status, Potentials and Further Work Need	2022	International Journal of Plant Production	(2022)
2	Gargiulo J.I., Lyons N.A., Clark C.E.F., Garcia S.C.	The AMS Integrated Management Model: A decision-support system for automatic milking systems	2022	Computers and Electronics in Agriculture	(2022)
3	Li H., Li S., Yu J., Han Y., Dong A.	AIoT Platform Design Based on Front and Rear End Separation Architecture for Smart Agricultural	2022	ACM International Conference Proceeding Series	(2022)
4	Kassanuk T., Phasinam K.	Impact of Internet of Things and Machine Learning in Smart Agriculture	2022	ECS Transactions
5	Ahamed N.N., Vignesh R.	Smart Agriculture and Food Industry with Blockchain and Artificial Intelligence	2022	Journal of Computer Science
6	Sood A., Sharma R.K., Bhardwaj A.K.	Artificial intelligence research in agriculture: a review	2022	Online Information Review	(2022)
7	Chiles R.M., Broad G., Gagnon M., Negowetti N., Glenna L., Griffin M.A.M., Tami-Barrera L., Baker S., Beck K.	Democratizing ownership and participation in the 4th Industrial Revolution: challenges and opportunities in cellular agriculture	2021	Agriculture and Human Values	(2021)
8	Mohr S., Kühl R.	Acceptance of artificial intelligence in German agriculture: an application of the technology acceptance model and the theory of planned behavior	2021	Precision Agriculture
9	Khan N., Kamaruddin M.A., Sheikh U.U., Yusup Y., Bakht M.P.	Oil palm and machine learning: Reviewing one decade of ideas, innovations, applications and gaps	2021	Agriculture (Switzerland)	(2021)
10	Bakhtadze N., Maximov E., Maximova N.	Local Wheat Price Prediction Models	2021	2021 7th International Conference on Control Science and Systems Engineering, ICCSSE 2021	(2021)
11	Eashwar S., Chawla P.	Evolution of Agritech Business 4.0 – Architecture and Future Research Directions	2021	IOP Conference Series: Earth and Environmental Science
12	Bogomolov A., Nevezhin V., Larionova M., Piskun E.	Review of digital technologies in agriculture as a factor that removes the growth limits to human civilization	2021	E3S Web of Conferences	(2021)
13	Wakjira K., Negera T., Zacepins A., Kviesis A., Komasilovs V., Fiedler S., Kirchner S., Hensel O., Purnomo D., Nawawi M., Paramita A., Rachman O.F., Pratama A., Faizah N.A., Lemma M., Schaedlich S., Zur A., Sper M., Proschek K., Gratzer K., Brodschneider R.	Smart apiculture management services for developing countries—the case of SAMS project in Ethiopia and Indonesia	2021	PeerJ Computer Science	(2021)
14	Panpatte S., Ganeshkumar C.	Artificial Intelligence in Agriculture Sector: Case Study of Blue River Technology	2021	Lecture Notes in Networks and Systems
15	Choi J., Koshizuka N.	Optimal Harvest date Prediction by Integrating Past and Future Feature Variables	2019	2019 IEEE Asia–Pacific Conference on Computer Science and Data Engineering, CSDE 2019
16	Backman J., Linkolehto R., Koistinen M., Nikander J., Ronkainen A., Kaivosoja J., Suomi P., Pesonen L.	Cropinfra research data collection platform for ISO 11783 compatible and retrofit farm equipment	2019	Computers and Electronics in Agriculture	(2019)
17	Thomas D.T., Mitchell P.J., Zurcher E.J., Herrmann N.I., Pasanen J., Sharman C., Henry D.A.	Pasture API: A digital platform to support grazing management for southern Australia	2019	23rd International Congress on Modelling and Simulation - Supporting Evidence-Based Decision Making: The Role of Modelling and Simulation, MODSIM 2019	(2019)
18	Skobelev P., Larukchin V., Mayorov I., Simonova E., Yalovenko O.	Smart Farming – Open Multi-agent Platform and Eco-System of Smart Services for Precision Farming	2019	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)	(2019)
19	Kamariotou M., Kitsios F., Madas M., Manthou V., Vlachopoulou M.	Strategic Decision Making and Information Management in the Agrifood Sector	2019	Communications in Computer and Information Science	(2019)
20	Sahu S., Chawla M., Khare N.	Viable crop prediction scenario in bigdata using a novel approach	2019	Advances in Intelligent Systems and Computing	(2019)
21	Balaji Prabhu B.V., Dakshayini M.	Performance Analysis of the Regression and Time Series Predictive Models using Parallel Implementation for Agricultural Data	2018	Procedia Computer Science
22	Rao M., Chhabria R., Gunasekaran A., Mandal P.	Improving competitiveness through performance evaluation using the APC model: A case in micro-irrigation	2018	International Journal of Production Economics	(2018)
23	Li J., Gao H., Liu Y.	Requirement analysis for the one-stop logistics management of fresh agricultural products	2017	Journal of Physics: Conference Series	(2017b)
24	Wolfert S., Ge L., Verdouw C., Bogaardt M.-J.	Big Data in Smart Farming – A review	2017	Agricultural Systems	(2017)
25	Nada A., Nasr M., Salah M.	Service oriented approach for decision support systems	2014	2014 IEEE 7th Joint International Information Technology and Artificial Intelligence Conference, ITAIC 2014	(2014)
26	Vizzari M., Modica G.	Environmental effectiveness of swine sewage management: A multicriteria AHP-based model for a reliable quick assessment	2013	Environmental Management
27	Lima M.L., Romanelli A., Massone H.E.	Decision support model for assessing aquifer pollution hazard and prioritizing groundwater resources management in the wet Pampa plain, Argentina	2013	Environmental Monitoring and Assessment	(2013)
28	Le Page M., Berjamy B., Fakir Y., Bourgin F., Jarlan L., Abourida A., Benrhanem M., Jacob G., Huber M., Sghrer F., Simonneaux V., Chehbouni G.	An Integrated DSS for Groundwater Management Based on Remote Sensing. The Case of a Semi-arid Aquifer in Morocco	2012	Water Resources Management	(2012)
29	Deng J., Chen X., Du Z., Zhang Y.	Soil Water Simulation and Predication Using Stochastic Models Based on LS-SVM for Red Soil Region of China	2011	Water Resources Management	(2011)
30	Carmona G., Varela-Ortega C., Bromley J.	The Use of Participatory Object-Oriented Bayesian Networks and Agro-Economic Models for Groundwater Management in Spain	2011	Water Resources Management	(2011)
31	Tironi A., Marin V.H., Campuzano F.J.	A management tool for assessing aquaculture environmental impacts in Chilean Patagonian fjords: Integrating hydrodynamic and pellets dispersion models	2010	Environmental Management	(2010)
32	Manos B.D., Papathanasiou J., Bournaris T., Voudouris K.	A DSS for sustainable development and environmental protection of agricultural regions	2010	Environmental Monitoring and Assessment	(2010)
33	d'Orgeval T., Boulanger J.-P., Capalbo M.J., Guevara E., Penalba O., Meira S.	Yield estimation and sowing date optimization based on seasonal climate information in the three CLARIS sites	2010	Climatic Change	(2010)
34	Wang H., Zhang X., Wang W., Zheng Y.	Research and implement of maize variety promotion decision support system based on WebGIS	2009	IFIP International Federation for Information Processing	(2009)
35	Nangia V., Turral H., Molden D.	Increasing water productivity with improved N fertilizer management	2008	Irrigation and Drainage Systems	(2008)
36	Cabrera V.E., Breuer N.E., Hildebrand P.E.	Participatory modeling in dairy farm systems: A method for building consensual environmental sustainability using seasonal climate forecasts	2008	Climatic Change	(2008)
37	Diaz B., Ribeiro A., Bueno R., Guinea D., Barroso J., Ruiz D., Fernadez-Quintanilla C.	Modelling wild-oat density in terms of soil factors: A machine learning approach	2005	Precision Agriculture	(2005)

Authors work

Category	Variables	Results	%
Authors		37
	Academics	25	67%
	Collaborations	8	22%
	Practitioners	4	11%
Type of source		37
	Article	21	57%
	Conference proceeding	16	43%
Location of the study		37
	Yes	24	65%
		11	46%
		7	29%
		6	24%
		2	8%
		2	8%
	No	13	35%
Research method		37
	Case study	26	70%
	Literature review	11	30%
Agricultural sector		37
	Cultivation of plants	15	40%
	General terms	15	40%
	Animal production	6	16%
	Fish farming	1	3%
Problems to solve-objective to achieve		37
	Increase efficiency and optimization maximizing farm returns	26	70%
	Manage the environmental impact and external changes	24	65%
	Predict and manage the farm complexity	19	51%
	Feed the increasing global population-food security	9	24%
	Other objectives	2	5%
Technology used		37
	Decision support system (DSS)	21	57%
	Artificial intelligence and machine learning	18	49%
	Big data analytics	16	43%
	Internet of things (IOT)	15	40%
	Drones	8	22%
	Robots	8	22%
	Cloud computing	7	19%
	Geographical indication system (GIS)	6	16%
	Other technologies	6	16%
	Biotechnology	4	11%
	Blockchain	3	8%
	Autonomous devices	3	8%
Applications in agriculture		37
	Precision farming and agronomic applications	24	65%
	Agronomic planning and economic applications	21	57%
	Water optimization and environmental management applications	15	40%
	Food supply chain applications and traceability	5	14%
Mentions a business model		37
	No	20	54%
	Yes	17	46%
		13	76%
		8	47%
		2	15%
Mentions the possibility to lead a new business model		37
	No	31	84%
	Yes	6	16%
		2	33%
		2	33%
		1	17%
		1	17%
Connects to sustainability issues		37
	Yes	23	62%
		8	35%
		6	26%
		5	22%
		5	22%
		4	17%
	No	14	38%
Explain the advantages		37
	Yes	34	92%
		24	71%
		16	47%
		2	6%
		2	6%
	No	3	8%
Explain the disadvantages		37
	No	30	81%
	Yes	7	19%
		1	14%
		1	14%
		1	14%
		1	14%
		1	14%
		1	14%
		1	14%
		1	14%
Explain the barriers		37
	No	23	62%
	Yes	14	38%
		7	50%
		7	50%
		6	43%
		4	29%
		3	21%
		2	14%
		2	14%
		1	7%
Research implications		37
	No	21	57%
	Yes	16	43%
		10	62%
		4	25%
		3	19%
		3	19%
Practical implications		37
	Yes	26	70%
		13	35%
		10	27%
		7	19%
		3	8%
	No	11	30%
Policy implications		37
	No	28	76%
	Yes	9	24%
		4	44%
		4	44%
		2	22%
		1	11%

Authors work

Problems to solve-Objectives to achieve		37
		26
		15
		12
		6
		4
		4
		24
		14
		8
		6
		4
		3
		1
		1
		19
		11
		9
		4
		2
		2
		9
		2
		1
		1

Authors work

Source(s): Authors work

Scopus advanced search string: “TITLE-ABS-KEY (artificial AND intelligence AND agriculture AND business AND model)”

Ahamed , N.N. and Vignesh , R. ( 2022 ), “ Smart agriculture and food industry with Blockchain and artificial intelligence ”, Journal of Computer Science , Vol.  18 No.  1 , doi: 10.3844/jcssp.2022.1.17 .

Ahmed , M. , Hayat , R. , Ahmad , M. , ul-Hassan , M. , Kheir , A.M.S. , ul-Hassan , F. , ur-Rehman , M.H. and Ahmad , S. ( 2022 ), “ Impact of climate change on dryland agricultural systems: a review of current status, potentials, and further work need ”, International Journal of Plant Production , Vol.  16 No.  3 , pp. 341 - 363 .

Amoussohoui , R. , Arouna , A. , Bavorova , M. , Tsangari , H. and Banout , J. ( 2022 ), “ An extended Canvas business model: a tool for sustainable technology transfer and adoption ”, Technology in Society , Vol.  68 , 101901 .

Arora , C. , Kamat , A. , Shanker , S. and Barve , A. ( 2022 ), “ Integrating agriculture and industry 4.0 under ‘agri-food 4.0’ to analyze suitable technologies to overcome agronomical barriers ”, British Food Journal , Vol.  124 No.  7 , pp.  2061 - 2095 .

Backman , J. , Linkolehto , R. , Koistinen , M. , Nikander , J. , Ronkainen , A. , Kaivosoja , J. , Suomi , P. and Pesonen , L. ( 2019 ), “ Cropinfra research data collection platform for ISO 11783 compatible and retrofit farm equipment ”, Computers and Electronics in Agriculture , Vol.  166 , 105008 .

Bagnoli , C. , Massaro , M. , Dal Mas , F. and Demartini , M. ( 2018 ), “ Defining the concept of Business Model. Searching for a business model framework ”, International Journal of Knowledge and Systems Science , Vol.  9 , pp.  48 - 64 .

Bagnoli , C. , Albarelli , A. , Biazzo , S. , Biotto , G. , Marseglia , G.R. , Massaro , M. , Messina , M. , Muraro , A. and Troiano , L. ( 2022 ), Digital Business Models for Industry 4.0 , Springer , Cham .

Bakhtadze , N. , Maximov , E. and Maximova , N. ( 2021 ), “ Local wheat price prediction models ”, 2021 IEEE 7th International Conference on Control Science and Systems Engineering (ICCSSE) , pp.  223 - 227 .

Balaji Prabhu , B.V. and Dakshayini , M. ( 2018 ), “ Performance analysis of the regression and time series predictive models using parallel implementation for agricultural data ”, Procedia Computer Science , Vol.  132 , pp.  198 - 207 .

Biancone , P.P. , Brescia , V. , Lanzalonga , F. and Alam , G.M. ( 2022 ), “ Using bibliometric analysis to map innovative business models for vertical farm entrepreneurs ”, British Food Journal , Vol.  124 No.  7 , pp.  2239 - 2261 .

Bogomolov , A. , Nevezhin , V. , Larionova , M. and Piskun , E. ( 2021 ), “ Review of digital technologies in agriculture as a factor that removes the growth limits to human civilization ”, E3S Web of Conferences , Vol.  247 , doi: 10.1051/e3sconf/202124701074 .

Boursianis , A.D. , Papadopoulou , M.S. , Diamantoulakis , P. , Liopa-Tsakalidi , A. , Barouchas , P. , Salahas , G. , Karagiannidis , G. , Wan , S. and Goudos , S.K. ( 2022 ), “ Internet of Things (IoT) and agricultural unmanned aerial vehicles (UAVs) in smart farming: a comprehensive review ”, Internet of Things , Vol.  18 , 100187 .

Bresciani , S. , Ciampi , F. , Meli , F. and Ferraris , A. ( 2021a ), “ Using big data for co-innovation processes: mapping the field of data-driven innovation, proposing theoretical developments and providing a research agenda ”, International Journal of Information Management , Vol.  60 , p.  102347 .

Bresciani , S. , Ferraris , A. and Del Giudice , M. ( 2018 ), “ The management of organizational ambidexterity through alliances in a new context of analysis: internet of Things (IoT) smart city projects ”, Technological Forecasting and Social Change , Vol.  136 , pp.  331 - 338 .

Bresciani , S. , Huarng , K.-H. , Malhotra , A. and Ferraris , A. ( 2021b ), “ Digital transformation as a springboard for product, process and business model innovation ”, Journal of Business Research , Vol.  128 , pp.  204 - 210 .

Cabrera , V.E. , Breuer , N.E. and Hildebrand , P.E. ( 2008 ), “ Participatory modeling in dairy farm systems: a method for building consensual environmental sustainability using seasonal climate forecasts ”, Climatic Change , Vol.  89 No.  3 , pp.  395 - 409 .

Carmona , G. , Varela-Ortega , C. and Bromley , J. ( 2011 ), “ The use of participatory object-oriented bayesian Networks and agro-economic models for groundwater management in Spain ”, Water Resources Management , Vol.  25 No.  5 , pp.  1509 - 1524 .

Chiles , R.M. , Broad , G. , Gagnon , M. , Negowetti , N. , Glenna , L. , Griffin , M.A.M. , Tami-Barrera , L. , Baker , S. and Beck , K. ( 2021 ), “ Democratizing ownership and participation in the 4th Industrial Revolution: challenges and opportunities in cellular agriculture ”, Agriculture and Human Values , Vol.  38 No.  4 , pp. 943 - 961 .

Choi , J. and Koshizuka , N. ( 2019 ), “ Optimal harvest date prediction by integrating past and future feature variables ”, 2019 IEEE Asia-Pacific Conference on Computer Science and Data Engineering (CSDE) , pp.  1 - 8 .

Chu , X. , Li , Y. , Tian , D. , Feng , J. and Mu , W. ( 2019 ), “ An optimized hybrid model based on artificial intelligence for grape price forecasting ”, British Food Journal , Vol.  121 No.  12 , pp.  3247 - 3265 .

Cobianchi , L. , Piccolo , D. , Dal Mas , F. , Agnoletti , V. , Ansaloni , L. , Balch , J. , Biffl , W. , Butturini , G. , Catena , F. , Coccolini , F. , Denicolai , S. , De Simone , B. , Frigerio , I. , Fugazzola , P. , Marseglia , G. , Marseglia , G.R. , Martellucci , J. , Modenese , M. , Previtali , P. , Ruta , F. , Venturi , A. , Kaafarani , H.M. and Loftus , T.J. , Team Dynamics Study Group ( 2023 ), “ Surgeons ’ perspectives on artificial intelligence to support clinical decision - making in trauma and emergency contexts : results from an international survey ”, World Journal of Emergency Surgery , Vol.  18 , pp. 1 - 17 .

Dal Mas , F. , Bagnoli , C. , Massaro , M. and Biazzo , S. ( 2021 ), “ Smart technologies and new business models: insights from artificial intelligence and Blockchain ”, in Chiucchi , M.S. , Lombardi , R. and Mancini , D. (Eds), Intellectual Capital, Smart Technologies and Digitalization : Emerging Issues and Opportunities , Springer International Publishing , Cham , pp.  271 - 285 .

Dal Mas , F. , Garcia-Perez , A. , Sousa , M.J. , Lopes da Costa , R. and Cobianchi , L. ( 2020 ), “ Knowledge translation in the healthcare sector. A structured literature review ”, Electronic Journal Of Knowledge Management , Vol.  18 No.  3 , pp.  198 - 211 .

Dal Mas , F. , Massaro , M. , Ndou , V. and Raguseo , E. ( 2023 ), “ Blockchain technologies for sustainability in the agrifood sector: a literature review of academic research and business perspectives ”, Technological Forecasting and Social Change , Vol.  187 , p.  122155 .

Deng , J. , Chen , X. , Du , Z. and Zhang , Y. ( 2011 ), “ Soil water simulation and predication using stochastic models based on LS-SVM for red soil region of China ”, Water Resources Management , Vol.  25 No.  11 , pp.  2823 - 2836 .

Diaz , B. , Ribeiro , A. , Bueno , R. , Guinea , D. , Barroso , J. , Ruiz , D. and Fernadez-Quintanilla , C. ( 2005 ), “ Modelling wild-oat density in terms of soil factors: a machine learning approach ”, Precision Agriculture , Vol.  6 No.  2 , pp.  213 - 228 .

DiVaio , A. , Palladino , R. , Hassan , R. and Escobar , O. ( 2020 ), “ Artificial intelligence and business models in the sustainable development goals perspective: a systematic literature review ”, Journal of Business Research , Vol.  121 , pp.  283 - 314 .

Doumpos , M. , Zopounidis , C. , Gounopoulos , D. , Platanakis , E. and Zhang , W. ( 2023 ), “ Operational research and artificial intelligence methods in banking ”, European Journal of Operational Research , Vol.  306 No.  1 , pp.  1 - 16 .

Dumay , J. , Bernardi , C. , Guthrie , J. and Demartini , P. ( 2016 ), “ Integrated reporting: a structured literature review ”, Accounting Forum , Vol.  40 No.  3 , pp.  166 - 185 .

d'Orgeval , T. , Boulanger , J.-P. , Capalbo , M.J. , Guevara , E. , Penalba , O. and Meira , S. ( 2010 ), “ Yield estimation and sowing date optimization based on seasonal climate information in the three CLARIS sites ”, Climatic Change , Vol.  98 No.  3 , pp.  565 - 580 .

Eashwar , S. and Chawla , P. ( 2021 ), “ Evolution of Agritech business 4.0 – architecture and future research directions ”, IOP Conference Series: Earth and Environmental Science , IOP Publishing , Vol.  775 No.  1 , p. 12011 .

European Commission ( 2017 ), “ Transforming our world: the 2030 agenda for sustainable development ”, available at: https://sdgs.un.org/2030agenda ( accessed 06 July 2023 ).

FAO ( 2022 ), “ World food and agriculture – statistical yearbook 2022 ”, Rome, available at: https://www.fao.org/documents/card/en/c/cc2211en

Gargiulo , J.I. , Lyons , N.A. , Clark , C.E.F. and Garcia , S.C. ( 2022 ), “ The AMS integrated management model: a decision-support system for automatic milking systems ”, Computers and Electronics in Agriculture , Vol.  196 , 106904 .

Goel , P. , Kaushik , N. , Sivathanu , B. , Pillai , R. and Vikas , J. ( 2022 ), “ Consumers' adoption of artificial intelligence and robotics in hospitality and tourism sector: literature review and future research agenda ”, Tourism Review , Vol.  77 No.  4 , pp.  1081 - 1096 .

GT staff reporters ( 2023 ), “ Chinese firms back to CES in Las Vegas, displaying latest technology ”, Global Times, available at: https://www.globaltimes.cn/page/202301/1283262.shtml

Kamariotou , M. , Kitsios , F. , Madas , M. , Manthou , V. and Vlachopoulou , M. ( 2019 ), “ Strategic decision making and information management in the agrifood sector BT - information and communication technologies in modern agricultural development ”, in Salampasis , M. and Bournaris , T. (Eds.), Springer International Publishing , Cham , pp.  97 - 109 .

Kassanuk , T. and Phasinam , K. ( 2022 ), “ Impact of internet of things and machine learning in smart agriculture ”, ECS Transactions , Vol.  107 No.  1 , p. 3215 .

Khan , N. , Kamaruddin , M.A. , Sheikh , U.U. , Yusup , Y. and Bakht , M.P. ( 2021 ), “ Oil palm and machine learning: reviewing one decade of ideas, innovations, applications, and gaps ”, Agriculture , Vol.  11 No.  9 , doi: 10.3390/agriculture11090832 .

Kok , J.N. , Boers , E.J.W. , Kosters , W.A. , Putten , P.V.D. and Poel , M. ( 2002 ), Artificial Intelligence: Definition, Trends, Techniques, and Cases , UNESCO , Paris .

Kraus , S. , Breier , M. and Dasí-Rodríguez , S. ( 2020 ), “ The art of crafting a systematic literature review in entrepreneurship research ”, International Entrepreneurship and Management Journal , Vol.  16 No.  3 , pp.  1023 - 1042 .

Kraus , S. , Breier , M. , Lim , W.M. , Dabić , M. , Kumar , S. , Kanbach , D. , Mukherjee , D. , Corvello , V. , Piñeiro-Chousa , J. , Liguori , E. , Palacios-Marqués , D. , Schiavone , F. , Ferraris , A. , Fernandes , C. and Ferreira , J.J. ( 2022 ), “ Literature reviews as independent studies: guidelines for academic practice ”, Review of Managerial Science , Vol.  16 No.  8 , pp. 2577 - 2595 .

Kulida , E. and Lebedev , V. ( 2020 ), “ About the use of artificial intelligence methods in aviation ”, 2020 13th International Conference “Management of Large-Scale System Development” (MLSD) , pp.  1 - 5 .

Le Page , M. , Berjamy , B. , Fakir , Y. , Bourgin , F. , Jarlan , L. , Abourida , A. , Benrhanem , M. , Jacob , G. , Huber , M. , Sghrer , F. , Simonneaux , V. and Chehbouni , G. ( 2012 ), “ An integrated DSS for groundwater management based on remote sensing. The case of a semi-arid aquifer in Morocco ”, Water Resources Management , Vol.  26 No.  11 , pp. 3209 - 3230 .

Lebelo , K. , Masinde , M. , Malebo , N. and Mochane , M.J. ( 2022 ), “ The surveillance and prediction of food contamination using intelligent systems: a bibliometric analysis ”, British Food Journal , Vol.  124 No.  4 , pp.  1149 - 1169 .

Li , J. , Gao , H. and Liu , Y. ( 2017 b), “ Requirement analysis for the one-stop logistics management of fresh agricultural products ”, Journal of Physics: Conference Series , Vol.  887 No.  1 , 12004 .

Li , B.H. , Hou , B.C. , Yu , W.T. , Lu , X.B. and Yang , C.W. ( 2017 a), “ Applications of artificial intelligence in intelligent manufacturing: a review ”, Frontiers of Information Technology and Electronic Engineering , Vol.  18 No.  1 , pp. 86 - 96 .

Li , H. , Li , S. , Yu , J. , Han , Y. and Dong , A. ( 2022 ), “ AIoT platform design based on front and rear end separation architecture for smart agricultural ”, 2022 4th Asia Pacific Information Technology Conference , New York, NY , Association for Computing Machinery , pp.  208 - 214 .

Lima , M.L. , Romanelli , A. and Massone , H.E. ( 2013 ), “ Decision support model for assessing aquifer pollution hazard and prioritizing groundwater resources management in the wet Pampa plain, Argentina ”, Environmental Monitoring and Assessment , Vol.  185 No.  6 , pp.  5125 - 5139 .

Loftus , T.J. , Tighe , P.J. , Filiberto , A.C. , Efron , P.A. , Brakenridge , S.C. , Mohr , A.M. , Rashidi , P. , Upchurch , G.R. , Jr. and Bihorac , A. ( 2020 ), “ Artificial intelligence and surgical decision-making ”, JAMA Surgery , Vol.  155 No.  2 , pp. 148 - 158 .

Magasumovna , A.A. , Anasovna , S.G. and Alfirovna , Z.Z. ( 2017 ), “ Factors and reserves of increase of efficiency of agricultural production ”, International Journal of Applied Engineering Research , Vol.  12 No.  24 , pp.  15821 - 15829 .

Manos , B.D. , Papathanasiou , J. , Bournaris , T. and Voudouris , K. ( 2010 ), “ A DSS for sustainable development and environmental protection of agricultural regions ”, Environmental Monitoring and Assessment , Vol.  164 No.  1 , pp.  43 - 52 .

Massaro , M. , Dumay , J. and Garlatti , A. ( 2015 ), “ Public sector knowledge management: a structured literature review ”, Journal of Knowledge Management , Vol.  19 No.  3 , pp.  530 - 558 .

Massaro , M. , Dumay , J.C. and Guthrie , J. ( 2016 ), “ On the shoulders of giants: undertaking a structured literature review in accounting ”, Accounting, Auditing and Accountability Journal , Vol.  29 No.  5 , pp.  767 - 901 .

Massaro , M. , Dumay , J. , Garlatti , A. and Dal Mas , F. ( 2018 ), “ Practitioners' views on intellectual capital and sustainability: from a performance-based to a worth-based perspective ”, Journal of Intellectual Capital , Vol.  19 No.  2 , pp.  367 - 386 .

Misra , N.N. , Dixit , Y. , Al-Mallahi , A. , Bhullar , M.S. , Upadhyay , R. and Martynenko , A. ( 2022 ), “ IoT, big data, and artificial intelligence in agriculture and food industry ”, IEEE Internet of Things Journal , Vol.  9 No.  9 , pp.  6305 - 6324 .

Mohr , S. and Kühl , R. ( 2021 ), “ Acceptance of artificial intelligence in German agriculture: an application of the technology acceptance model and the theory of planned behavior ”, Precision Agriculture , Vol.  22 No.  6 , pp.  1816 - 1844 .

Musa , S.F.P.D. and Basir , K.H. ( 2021 ), “ Smart farming: towards a sustainable agri-food system ”, British Food Journal , Vol.  123 No.  9 , pp.  3085 - 3099 .

Nada , A. , Nasr , M. and Salah , M. ( 2014 ), “ Service oriented approach for decision support systems ”, 2014 IEEE 7th Joint International Information Technology and Artificial Intelligence Conference , pp.  409 - 413 .

Nangia , V. , Turral , H. and Molden , D. ( 2008 ), “ Increasing water productivity with improved N fertilizer management ”, Irrigation and Drainage Systems , Vol.  22 No.  3 , pp.  193 - 207 .

Page , M.J. , McKenzie , J.E. , Bossuyt , P.M. , Boutron , I. , Hoffmann , T.C. , Mulrow , C.D. , Shamseer , L. , Tetzlaff , J.M. , Akl , E.A. , Brennan , S.E. , Chou , R. , Glanville , J. , Grimshaw , J.M. , Hróbjartsson , A. , Lalu , M.M. , Li , T. , Loder , E.W. , Mayo-Wilson , E. , McDonald , S. , McGuinness , L.A. , Stewart , L.A. , Thomas , J. , Tricco , A.C. , Welch , V.A. , Whiting , P. and Moher , D. ( 2021 ), “ The PRISMA 2020 statement: an updated guideline for reporting systematic reviews ”, BMJ , Vol.  372 No.  71 , doi: 10.1136/bmj.n71 .

Panpatte , S. and Ganeshkumar , C. ( 2021 ), “ Artificial intelligence in agriculture sector: case study of blue river technology BT - Proceedings of the second international conference on information management and machine intelligence ”, in Goyal , D. , Gupta , A.K. , Piuri , V. , Ganzha , M. and Paprzycki , M. (Eds), Springer Singapore , Singapore , pp.  147 - 153 .

Paoloni , P. , Modaffari , G. and Mattei , G. ( 2021 ), “ Knowledge resources in the university context: an overview of the literature ”, Journal of Intellectual Capital , Vol.  22 No.  4 , pp.  703 - 724 .

Paul , J. , Lim , W.M. , O'Cass , A. , Hao , A.W. and Bresciani , S. ( 2021 ), “ Scientific procedures and rationales for systematic literature reviews (SPAR-4-SLR) ”, International Journal of Consumer Studies , Vol.  45 No.  4 , pp.  O1 - O16 .

Pietrewicz , L. ( 2019 ), “ Technology, business models and competitive advantage in the age of industry 4.0 ”, Problemy Zarzadzania , Vol.  82 , pp.  32 - 52 .

Rao , M. , Chhabria , R. , Gunasekaran , A. and Mandal , P. ( 2018 ), “ Improving competitiveness through performance evaluation using the APC model: a case in micro-irrigation ”, International Journal of Production Economics , Vol.  195 , pp.  1 - 11 .

Romanello , R. and Veglio , V. ( 2022 ), “ Industry 4.0 in food processing: drivers, challenges and outcomes ”, British Food Journal , Vol.  124 No.  13 , pp.  375 - 390 .

Saad , M.H. , Hamdan , N.M. and Sarker , M.R. ( 2021 ), “ State of the art of urban smart vertical farming automation system: advanced topologies, issues and recommendations ”, Electronics , Vol.  10 No.  12 , doi: 10.3390/electronics10121422 .

Sahu , S. , Chawla , M. and Khare , N. ( 2019 ), “ Viable crop prediction scenario in BigData using a novel approach BT - emerging technologies in data mining and information security ”, in Abraham , A. , Dutta , P. , Mandal , J.K. , Bhattacharya , A. and Dutta , S. (Eds), Emerging Technologies in Data Mining and Information Security. Advances in Intelligent Systems and Computing , Springer Singapore , Singapore , pp.  165 - 177 .

Schünemann , H. , Higgins , J. , Vist , G. , Glasziou , P. , Akl , E. , Skoetz , N. and Guyatt , G. ( 2021 ), “ Completing ‘summary of findings’ tables and grading the certainty of the evidencele ”, in Cochrane , Higgins , J.P.T. , Thomas , J. , Chandler , J. , Cumpston , M. , Li , T. , Page , M.J. and Welch , V.A. (Eds), Cochrane Handbook for Systematic Reviews of Interventions Version 6.2 (Updated February 2021) , Chapter 14 .

Secinaro , S. and Calandra , D. ( 2021 ), “ Halal food: structured literature review and research agenda ”, British Food Journal , Vol.  123 No.  1 , pp.  225 - 243 .

Secinaro , S. , Dal Mas , F. , Massaro , M. and Calandra , D. ( 2022 ), “ Exploring agricultural entrepreneurship and new technologies: academic and practitioners' views ”, British Food Journal , Vol.  124 No.  7 , pp.  2096 - 2113 .

Shrestha , Y.R. , Ben-Menahem , S.M. and von Krogh , G. ( 2019 ), “ Organizational decision-making structures in the age of artificial intelligence ”, California Management Review , Vol.  61 No.  4 , pp.  66 - 83 .

Shukla , S. and Sengupta , T. ( 2021 ), “ Business model innovation in the agricultural supply chain at Bottom of the Pyramid: evidence from India ”, Strategic Change , Vol.  30 No.  5 , pp.  461 - 466 .

Skobelev , P. , Larukchin , V. , Mayorov , I. , Simonova , E. and Yalovenko , O. ( 2019 ), “ Smart farming – open multi-agent platform and eco-system of smart services for precision farming BT - advances in practical applications of survivable agents and multi-agent systems: the PAAMS collection ”, in Demazeau , Y. , Matson , E. , Corchado , J.M. and De la Prieta , F. (Eds), Advances in Practical Applications of Survivable Agents and Multi-Agent Systems: The PAAMS Collection. PAAMS 2019. Lecture Notes in Computer Science , Springer International Publishing , Cham , pp.  212 - 224 .

Sood , A. , Sharma , R.K. and Bhardwaj , A.K. ( 2022 ), “ Artificial intelligence research in agriculture: a review ”, Online Information Review , Vol.  46 No.  6 , pp.  1054 - 1075 .

Sunding , D. and Zilbermanof , D.B.T.-H.A.E. ( 2001 ), “ Chapter 4 the agricultural innovation process: research and technology adoption in a changing agricultural sector ”, Agricultural Production , Vol.  1 , pp.  207 - 261 .

Thomas , D.T. , Mitchell , P.J. , Zurcher , E.J. , Herrmann , N.I. , Pasanen , J. , Sharman , C. and Henry , D.A. ( 2019 ), “ Pasture API: a digital platform to support grazing management for southern Australia ”, 23rd International Congress on Modelling and Simulation , Camberra , pp.  393 - 399 .

Tironi , A. , Marin , V.H. and Campuzano , F.J. ( 2010 ), “ A management tool for assessing aquaculture environmental impacts in Chilean Patagonian Fjords: integrating hydrodynamic and pellets dispersion models ”, Environmental Management , Vol.  45 No.  5 , pp.  953 - 962 .

Trivelli , L. , Apicella , A. , Chiarello , F. , Rana , R. , Fantoni , G. and Tarabella , A. ( 2019 ), “ From precision agriculture to Industry 4.0 ”, British Food Journal , Vol.  121 No.  8 , pp.  1730 - 1743 .

UN Environment Programme ( 2021 ), “ UNEP food waste index report 2021 ”, available at: https://www.unep.org/resources/report/unep-food-waste-index-report-2021

VanGerrewey , T. , Boon , N. and Geelen , D. ( 2022 ), “ Vertical farming: the only way is up? ”, Agronomy , Vol.  12 No.  1 , doi: 10.3390/agronomy12010002 .

Vaska , S. , Massaro , M. , Bagarotto , E.M. and Dal Mas , F. ( 2021 ), “ The digital transformation of business model innovation: a structured literature review ”, Frontiers in Psychology , Vol.  11 , p. 3557 .

Vizzari , M. and Modica , G. ( 2013 ), “ Environmental effectiveness of swine sewage management: a multicriteria AHP-based model for a reliable quick assessment ”, Environmental Management , Vol.  52 No.  4 , pp.  1023 - 1039 .

Wakjira , K. , Negera , T. , Zacepins , A. , Kviesis , A. , Komasilovs , V. , Fiedler , S. , Kirchner , S. , Hensel , O. , Purnomo , D. , Nawawi , M. , Paramita , A. , Fauzi Rachman , O. , Pratama , A. , Al Faizah , N. , Lemma , M. , Schaedlich , S. , Zur , A. , Sperl , M. , Proschek , K. , Gratzer , K. and Brodschneider , R. ( 2021 ), “ Smart apiculture management services for developing countries—the case of SAMS project in Ethiopia and Indonesia ”, in Procter , J. (Ed.), PeerJ Computer Science , PeerJ , Vol.  7 , p. e484 .

Wamba-Taguimdje , S.L. , Fosso Wamba , S. , Kala Kamdjoug Jean , R. and Tchatchouang Wanko , C.E. ( 2020 ), “ Influence of artificial intelligence (AI) on firm performance: the business value of AI-based transformation projects ”, Business Process Management Journal , Vol.  26 No.  7 , pp.  1893 - 1924 .

Wang , H. , Zhang , X. , Wang , W. and Zheng , Y. ( 2009 ), “ Research and implement of maize variety promotion decision support system based onwebgis BT - computer and computing technologies in agriculture II, volume 1 ”, in Li , D. and Zhao , C. (Eds), Computer and Computing Technologies in Agriculture II, Volume 1. CCTA 2008. IFIP Advances in Information and Communication Technology , Springer US , Boston, MA , pp. 715 - 723 .

Wijerathna-Yapa , A. and Pathirana , R. ( 2022 ), “ Sustainable agro-food systems for addressing climate change and food security ”, Agriculture , Vol.  12 No.  10 , doi: 10.3390/agriculture12101554 .

Wolfert , S. , Ge , L. , Verdouw , C. and Bogaardt , M.-J. ( 2017 ), “ Big data in smart farming – a review ”, Agricultural Systems , Vol.  153 , pp.  69 - 80 .

Yela Aránega , A. , Ferraris , A. , Baima , G. and Bresciani , S. ( 2022 ), “ Guest editorial: sustainable growth and development in the food and beverage sector ”, British Food Journal , Vol.  124 No.  8 , pp.  2429 - 2433 .

Corresponding author

Related articles, all feedback is valuable.

Please share your general feedback

Report an issue or find answers to frequently asked questions

Contact Customer Support

• Publications
• Partnerships

Is there a potential in adopting Artificial Intelligence in food and agriculture sector, and can it transform food systems and with what impact?

By Thembani Malapela

• AI and Machine Learning technologies could transform the agri-food systems and help ending hunger.
• UN-Wide System is adopting the Artificial Intelligence in its programmes.
• The Food and Agriculture Organization of the United Nations (FAO) is adopting and experimenting with AI and Machine Learning across the agriculture domain.
• FAO is also aware of the impact of AI and is now a signatory to the " Rome Call for AI Ethics "

Introduction

In 2017, l wrote a blog on whether Artificial Intelligence (AI) can improve agricultural productivit y and at that time, the question was befitting, as this technology was still not yet widely appreciated. However, three years have passed, and the use and adoption of AI has grown in many sectors. Additionally, Covid-19 has forced governments, companies and individuals to rely heavily on digital technologies.

Attention is increasing on the potential of AI and Machine Learning (ML) to the Food and Agriculture sector, especially on transforming food systems and application to different value-chains. Recently, the International Finance Corporation (IFC) published a report on Artificial Intelligence in Emerging markets , and regarding agribusiness, it noted that, “Artificial intelligence can spur progress toward meeting Sustainable Development Goal #2 —to end hunger, achieve food security, improve nutrition, and promote sustainable agriculture” (p.72). The increased access to faster internet connection(s), investment in connectivity infrastructure and wider availability of smart phones (and other handheld devises) provides a rock bed for the uptake of digital technologies and especially AI amongst farmers even in developing countries. (Read example on M-Shwari Case Study )

Meanwhile, across the UN system-wide, the adoption of AI is well documented by the International Telecommunications Union (ITU) in a report that highlighted the use cases of AI amongst UN Agencies. This year the “ AI for Social Good Summit ” is pencilled for 21-30 September 2020. The summit is organised by the ITU and the XPRIZE Foundation, in partnership with UN sister agencies, Switzerland, and ACM. It offers a platform for discussing various AI related issues and application.

In authoring this piece, I have no intention to be an authority in this technology nor in agri-food systems; however, from a knowledge sharing perspective, l would like to disseminate some initiatives ( known to me at this time) on the use of AI and Machine Learning in food and agriculture by the Food and Agriculture Organization of the United Nations * . The blog will conclude with some issues arising in the adoption AI. 

FAO on Artificial Intelligence

Some snippets of publication of issues related to AI in FAO started appearing from 2019, initially with the launch of a mobile app to monitor fall armyworm. The availability of farm data increased and paved the way to develop and deploy AI in agriculture. The trend has also been spurred by major ag input companies, equipment manufacturers, and service providers who produced products and services that either consumed or diffused farm level data ( Rakestraw & Acharya, 2017 ). Furthermore, the development of AI algorithms and AI applications for the value chains have placed the private sector as a trendsetter in the digital agriculture ecosystem.

FAO China, in 2019 hosted a dialogue in partnership with the Chinese Academy of Agricultural Planning and Engineering (AAPE) themed the “2019 Dialogue on the Application of AI in Agriculture”. The dialogue gathered around 60 representatives from government, institutions, academia and private sector in Beijing ( FAO China ). That meeting offered insights from public sector, and founders of leading enterprises in the industry, and participants shared their practical experiences, cutting-edge technologies and concerns with AI adoption.

Recent examples on AI and Machine Learning in FAO

Some of the activities and projects where FAO has adopted the AI and Machine Learning in agriculture are listed below. This by no means an exhaustive list but representative of the few l managed to glean during the brief research and rightfully so have specific project leads within the organization.

• Fish Species Identification .

Fish species identification is classification and categorizing fish, by fisheries taxonomic experts, based on external morphological features, including body shape, pattern of colors, scale size and count, number and relative position of fins, number and type of fin rays, or various relative measurements of body parts ( Strauss and Bond,1990 ).

While a number of tools and guides (including web resources such as Fish Base and the Catalog of Fishes) were developed by FAO to support taxonomic experts, latest technologies have provided additional ease to this process.FAO has explored the use of Artificial Intelligence in Fish Species Identification by using Google Cloud AutoML. Cloud AutoML is a suite of machine learning products that enables developers with limited machine learning expertise to train high-quality models specific to their business needs. It relies on Google’s state-of-the-art transfer learning and neural architecture search technology. After testing different approaches, the FAO CSI team concluded using Google AutoML performed better and was quicker to develop that building the models ourselves even when starting with massively pre-trained models. The team discovered that good quality data was needed and a good mark-up workflow.

• 1b.       Species recognition

iSharkFin is an expert system that uses machine learning techniques to identify shark species from shark fin shapes. Aimed at port inspectors, custom agents, fish traders and other users without formal taxonomic training, iSharkFin allows the identification of shark species from a picture of the fin. The iSharkFin takes an interactive process.  Users only need to take a standard photo, select some characteristics of a fin and choose a few points on the fin shape, iSharkFin will automatically analyze the information and tell you the shark specie from which the fin comes ( Strauss and Bond,1990 )

• Advanced Land Monitoring

The FAO system for earth observations, data access, processing & analysis for land monitoring ( SEPA L) helps countries measure, monitor and report on forests and land use, offering unparalleled access to granular satellite data and computing power, for improved climate change mitigation plans and better-informed land-use policies. It uses advanced Cloud computing, AI and machine learning to provide comprehensive image processing capabilities and enables detection of small-scale changes in forests, such as those associated with illegal or unsustainable timber harvesting. The system provides the capability to countries to develop robust national forest monitoring systems, detecting Forest Degradation and Forest Fires. ( Bravi, 2019 )

• The FAO Geospatial Machine Learning pipeline: Extracting objects of interest from remote sensing imagery

Another exploratory work was using ML to extracts objects of interest from remote sensing imagery via a GeoML pipeline. More specifically, the machine was trained to detect palm trees (with some hints from Penn State on optimization). The approach was applied to fish cages and fish drying racks and vessels . A refinement was counting the identified vessels. The GeoML pipeline was deployed on Google ML and standalone for desktop use. See presentation here

• Management of scarce resources

Artificial Intelligence is used by FAO to make better use of scarce resources like water and energy. Achieving Food Security in the future while using water resources in a sustainable manner is a major challenge for our and next generations. Agriculture is a key water user. A careful monitoring of water productivity in agriculture and exploring opportunities to increase it are required. FAO has developed a publicly accessible near real time database using satellite data that allows monitoring of agricultural water productivity and do advanced water management (Read about WaPOR ) 

• Plant pest detection

Fall Armyworm is spreading fast across many parts of the world, including sub-Saharan Africa, devastating crops and farmers’ livelihoods (F AO,2018 ). A mobile phone application called FAMEWS , which uses machine learning and artificial intelligence, offers hope in tackling the pest problem. [2] Farmers can easily detect Fall Army Worm damage by using mobile phones. This augments the human intelligence and serves extra extension capacity. At the moment almost 20 plant pests can be detected already by using a mobile phone. [ Listen to the podcast here]

• Detection of Agriculture Stress

FAO has developed the Agricultural Stress Index (ASIS), a quick-look indicator for the early identification of agricultural areas probably affected by dry spells, or drought in extreme cases. It monitors agricultural areas with a high likelihood of water stress/drought at global, regional and country level, using satellite technology. Drought affects more people than any other type of natural disaster and is the most damaging to livelihoods, especially in developing countries.

• Support to investment

Earthmap is a new FAO tool based on Google Earth Engine providing simplified access to complex datasets. It facilitates the understanding of land cover and land use dynamics processes for designing and assessing baselines, monitoring and evaluation. (Read the emerging news item )

Potential brought by these technologies and issues arising

The food and agriculture sector still faces the inherent challenge of feed the ever-growing world’s population. Additionally, over 820 million people go hungry, around 2 billion more lack sufficient micronutrients and another 2.5 billion consume excess calories for their needs.

Through innovation, technology adoption and mechanization the agri-food systems has managed to survive until now, with a growing fear of whether amidst mounting global challenges (such as climate change, political unrests, etc) would the food systems cope with increasing food demands. The adoption of technologies, such as AI and Machine Leaning, could improve agricultural productivity.

There are claims that claims that AI capabilities could someday exceed human capabilities. As a result, there is a growing call that these new technologies should be researched and be produced in a way that they do not interfer with human rights, are environmentally friendly and thus not marginalizing the poor and most vulnerable. In this vein, FAO is one of the signatories for the “ Rome Call for AI Ethics"

Rome Call for AI Ethics – This is ethical resolution on Artificial Intelligence (AI) that stress the importance of minimizing this technology's risks while exploiting its potential benefits. The Rome Call for AI Ethics refers to the need for a highly sustainable approach, which also includes the use of artificial intelligence in insuring sustainable food systems in the future. ( FAO, 2020 ).

The future of Artificial Intelligence seems to be bright within the food and agriculture sector and with a likehood of transforming agri-food systems if upscaled homogeniously. FAO has adopted this technology in the areas highlighted above and the signatory to the Rome Call for AI Ethics affirms FAOs commitment to adopting sustainable technologies with a consideration to the human rights and rights of the poor and the marginalized.

Disclaimer:  This piece is my opinion piece based on the published reports and news and does not attempt to express the views, or represent the views and position of the Food and Agriculture Organization of the United Nations.

Topics: 

• Technologies
• cloud computing
• machine to machine
• sensor networks
• policy implementation
• stakeholders

Post your comment

More From Forbes

The future of farming: ai innovations that are transforming agriculture.

• Share to Facebook
• Share to Twitter
• Share to Linkedin

AI-assisted Agriculture

Agriculture is a cornerstone of human civilization, a testament to our ability to harness nature for sustenance. Yet, this age-old industry faces many challenges that hamper productivity, impact livelihoods, and threaten global food security.

By 2050, we must produce 60 percent more food to feed a world population of 9.3 billion, reports the Food and Agriculture Organization. Given the current industry challenges, doing that with a farming-as-usual approach could be tricky. Moreover, this would extend the heavy toll we already place on our natural resources.

This is where Artificial Intelligence can come to our rescue. The AI in Agriculture Market is projected to grow from $1.7 billion in 2023 to $4.7 billion by 2028, highlighting the pivotal role of advanced technologies in this sector. This article explores three significant issues agriculture faces today and shows how AI is helping tackle them using real-world examples.

Three key challenges farmers face

Amongst the many issues hurting farmers, three stand out due to their global presence and financial impact:

1. Pests : Pests devour approximately 40% of global agricultural productivity annually, costing at least $70 billion. From locust swarms decimating fields in Africa to fruit flies affecting orchards, the impact is global, and financial repercussions are colossal.

Best High-Yield Savings Accounts Of 2024

Best 5% interest savings accounts of 2024.

2. Soil Quality and Irrigation : Soil degradation affects nearly 33% of the Earth's soil, diminishing its ability to grow crops, leading to a loss of about $400 billion. Water scarcity and inefficient irrigation further dent agricultural output. Agriculture uses 70% of the world's accessible freshwater, but 60% of it is wasted due to leaky irrigation systems.

3. Weeds : Despite advancements in agricultural practices, weeds cause significant declines in crop yield and quality. Around 1800 weed species reduce plant production by about 31.5%, leading to economic losses of about $32 billion annually.

How AI is transforming Agriculture

Smart Farming

Artificial Intelligence is often used as a catchall phrase. Here, it refers to the systematic collection of data, pertinent use of analytics ranging from simple descriptive summaries to deep learning algorithms, and advanced technologies such as computer vision, the internet of things, and geospatial analytics. Let’s look at how AI helps address each of the above challenges:

1. Pest identification and control : Accurate, early identification and control of pests is essential to minimize crop damage and reduce the reliance on chemical pesticides. Data such as weather reports, historical pest activity, and high-resolution images captured by drones or satellites are readily available today. Machine learning models and computer vision can help predict pest invasions and identify pests in the field.

For example, Trapview has built a device that traps pests and identifies them. It uses pheromones to attract pests, which are photographed by a camera in the device. By leveraging Trapview’s database, AI identifies over 60 pest species, such as the codling moth, which afflicts apples, and the cotton bollworm, which can damage lettuce and tomatoes.

Once identified, the system uses location and weather data to map out the likely impact of the insects and pushes the findings as an app notification to farmers. These AI-driven insights enable timely and targeted interventions, significantly reducing crop losses and chemical usage. Trapview reports that its customers have seen a 5% increase in yield and quality, and overall savings of 118 million euro in growers’ costs.

2. Soil health monitoring : Continuous monitoring and analysis of soil health are essential to ensuring optimal growing conditions and sustainable farming practices. Optimizing water use is crucial to ensuring crops receive precisely what they need, reducing waste and enhancing productivity.

Data from in-ground sensors, farm machinery, drones, and satellites are used to analyze soil conditions, including moisture content, nutrient levels, and the presence of pathogens. Such soil health analysis helps predict water needs and automate irrigation systems.

For example, CropX has built a platform specializing in soil health monitoring by leveraging real-time data to help users review and compare vital parameters alongside crop performance. Farmers gain insights into soil type and vegetation indices like NDVI - normalized difference vegetation index, SAVI - soil adjusted vegetation index, and soil moisture index to optimize crop management strategies. CropX reports that its solutions have led to a 57% reduction in water usage, a 15% reduction in fertilizer usage, and up to 70% yield increase.

3. Weed Detection and Management : Precise identification and elimination of weeds is critical to preventing them from competing for precious resources with crops and minimizing herbicide use. Thanks to computer vision, drones and robots can now identify weeds amongst crops with high precision. This allows for targeted weed control, either mechanically or through precise herbicide application.

For instance, the startup Carbon Robotics leverages deep learning algorithms in its computer vision solution. It identifies weeds by analyzing data from over 42 high-resolution cameras that scan the fields in real-time. Then, it employs robotics and lasers to deliver high-precision weed control.

The LaserWeeder claims to weed up to two acres per hour and eliminate up to 5,000 weeds per minute at 99% accuracy. Its growers report reducing weed control costs by up to 80% with a potential return on investment in one to three years.

Tackling the risks of automation

Opportunities and risks of AI in agriculture

AI has numerous benefits for agriculture but isn’t without inherent risks , such as job displacement, ownership concentration, and ethical concerns. When AI automates tasks traditionally done by humans in large numbers, it could lead to job losses across both manual and cognitive roles. Moreover, it could exacerbate ownership concentration, benefiting large enterprises or wealthy individuals at the expense of smaller farms.

When farmland turns into a hotbed for data collection – underground, at the crop level, and from the sky, this could lead to data privacy issues. These challenges underscore the need for careful consideration and governance to balance AI's advantages against its potential downsides. This is unique not just to the agricultural sector but to all industries where AI is being applied.

Ushering in a transformative future

Integrating AI in agriculture is not just reshaping current practices but also paving the way for a sustainable and resilient future. AI could become a master gardener, perpetually monitoring and fine-tuning every growth stage in the farm, from seed selection to harvest and beyond. It can help adjust farming practices in real time to climatic shifts, ensuring optimal crop health and yield.

• Editorial Standards
• Reprints & Permissions

Join The Conversation

One Community. Many Voices. Create a free account to share your thoughts. 

Forbes Community Guidelines

Our community is about connecting people through open and thoughtful conversations. We want our readers to share their views and exchange ideas and facts in a safe space.

In order to do so, please follow the posting rules in our site's  Terms of Service.   We've summarized some of those key rules below. Simply put, keep it civil.

Your post will be rejected if we notice that it seems to contain:

• False or intentionally out-of-context or misleading information
• Insults, profanity, incoherent, obscene or inflammatory language or threats of any kind
• Attacks on the identity of other commenters or the article's author
• Content that otherwise violates our site's  terms.

User accounts will be blocked if we notice or believe that users are engaged in:

• Continuous attempts to re-post comments that have been previously moderated/rejected
• Racist, sexist, homophobic or other discriminatory comments
• Attempts or tactics that put the site security at risk
• Actions that otherwise violate our site's  terms.

So, how can you be a power user?

• Stay on topic and share your insights
• Feel free to be clear and thoughtful to get your point across
• ‘Like’ or ‘Dislike’ to show your point of view.
• Protect your community.
• Use the report tool to alert us when someone breaks the rules.

Thanks for reading our community guidelines. Please read the full list of posting rules found in our site's  Terms of Service.

Information

• Author Services

Initiatives

You are accessing a machine-readable page. In order to be human-readable, please install an RSS reader.

All articles published by MDPI are made immediately available worldwide under an open access license. No special permission is required to reuse all or part of the article published by MDPI, including figures and tables. For articles published under an open access Creative Common CC BY license, any part of the article may be reused without permission provided that the original article is clearly cited. For more information, please refer to https://www.mdpi.com/openaccess .

Feature papers represent the most advanced research with significant potential for high impact in the field. A Feature Paper should be a substantial original Article that involves several techniques or approaches, provides an outlook for future research directions and describes possible research applications.

Feature papers are submitted upon individual invitation or recommendation by the scientific editors and must receive positive feedback from the reviewers.

Editor’s Choice articles are based on recommendations by the scientific editors of MDPI journals from around the world. Editors select a small number of articles recently published in the journal that they believe will be particularly interesting to readers, or important in the respective research area. The aim is to provide a snapshot of some of the most exciting work published in the various research areas of the journal.

Original Submission Date Received: .

• Active Journals
• Find a Journal
• Proceedings Series
• For Authors
• For Reviewers
• For Editors
• For Librarians
• For Publishers
• For Societies
• For Conference Organizers
• Open Access Policy
• Institutional Open Access Program
• Special Issues Guidelines
• Editorial Process
• Research and Publication Ethics
• Article Processing Charges
• Testimonials
• Preprints.org
• SciProfiles
• Encyclopedia

Article Menu

• Subscribe SciFeed
• Recommended Articles
• Google Scholar
• on Google Scholar
• Table of Contents

Find support for a specific problem in the support section of our website.

Please let us know what you think of our products and services.

Visit our dedicated information section to learn more about MDPI.

JSmol Viewer

Utilizing artificial intelligence techniques for a long–term water resource assessment in the shihmen reservoir for water resource allocation.

1. Introduction

2. materials and methods, 2.1. study area, 2.2. methodology construction, 2.2.1. simulation model of long–term water resource supply, 2.2.2. long–term water resource demand assessment model, 2.2.3. model evaluation, 3. results and discussion, 3.1. the performance of the long–term water resource supply simulation, 3.2. the performance of the long–term water resource demand estimation, 3.3. integration with decision support systems (dsss).

• Data Integration: The DSS interface feeds real–time data from meteorological stations and relevant sources into the models.
• Model Execution: The DSS triggers the models to run using the latest data inputs, generating updated forecasts.
• Output Interpretation: The DSS interprets the model outputs, including predictions of inflow and water demand, to inform water management strategies.
• Decision–Making Support: The DSS utilizes the model outputs to suggest or automate decisions, such as reservoir release schedules or water rationing measures.

4. Conclusions

5. future recommendations, author contributions, data availability statement, acknowledgments, conflicts of interest.

• Bai, Y.; Chen, Z.; Xie, J.; Li, C. Daily Reservoir Inflow Forecasting Using Multiscale Deep Feature Learning with Hybrid Models. J. Hydrol. 2016 , 532 , 193–206. [ Google Scholar ] [ CrossRef ]
• Sheikh, M.R.; Coulibaly, P. Review of Recent Developments in Hydrologic Forecast Merging Techniques. Water 2024 , 16 , 301. [ Google Scholar ] [ CrossRef ]
• Nakhaei, M.; Zanjanian, H.; Nakhaei, P.; Gheibi, M.; Moezzi, R.; Behzadian, K.; Campos, L.C. Comparative Evaluation of Deep Learning Techniques in Streamflow Monthly Prediction of the Zarrine River Basin. Water 2024 , 16 , 208. [ Google Scholar ] [ CrossRef ]
• Yan, K.; Gao, S.; Wen, J.; Yao, S. A Multi–Factor Combination Model for Medium to Long–Term Runoff Prediction Based on Improved BP Neural Network. Water 2023 , 15 , 3559. [ Google Scholar ] [ CrossRef ]
• Zhang, F.; Wu, Z.; Di, D.; Wang, H. Water Resources Allocation Based on Water Resources Supply–Demand Forecast and Comprehensive Values of Water Resources. J. Hydrol. Reg. Stud. 2023 , 47 , 101421. [ Google Scholar ] [ CrossRef ]
• Tian, J.; Guo, S.; Deng, L.; Yin, J.; Pan, Z.; He, S.; Li, Q. Adaptive Optimal Allocation of Water Resources Response to Future Water Availability and Water Demand in the Han River Basin, China. Sci. Rep. 2021 , 11 , 7879. [ Google Scholar ] [ CrossRef ]
• Chang, F.-J.; Wang, Y.-C.; Tsai, W.-P. Modelling Intelligent Water Resources Allocation for Multi-Users. Water Resour. Manag. 2016 , 30 , 1395–1413. [ Google Scholar ] [ CrossRef ]
• Kim, S.; Kim, B.S.; Jun, H.; Kim, H.S. Assessment of Future Water Resources and Water Scarcity Considering the Factors of Climate Change and Social–Environmental Change in Han River Basin, Korea. Stoch. Environ. Res. Risk Assess. 2014 , 28 , 1999–2014. [ Google Scholar ] [ CrossRef ]
• Ghalehkhondabi, I.; Ardjmand, E.; Young, W.A.; Weckman, G.R. Water Demand Forecasting: Review of Soft Computing Methods. Environ. Monit. Assess. 2017 , 189 , 313. [ Google Scholar ] [ CrossRef ]
• Niknam, A.; Zare, H.K.; Hosseininasab, H.; Mostafaeipour, A.; Herrera, M. A Critical Review of Short-Term Water Demand Forecasting Tools—What Method Should I Use? Sustainability 2022 , 14 , 5412. [ Google Scholar ] [ CrossRef ]
• Khazaee Poul, A.; Shourian, M.; Ebrahimi, H. A Comparative Study of MLR, KNN, ANN and ANFIS Models with Wavelet Transform in Monthly Stream Flow Prediction. Water Resour. Manag. 2019 , 33 , 2907–2923. [ Google Scholar ] [ CrossRef ]
• Kim, D.; Choi, S.; Kang, S.; Noh, H. A Study on Developing an AI-Based Water Demand Prediction and Classification Model for Gurye Intake Station. Water 2023 , 15 , 4160. [ Google Scholar ] [ CrossRef ]
• Msiza, I.S.; Nelwamondo, F.V.; Marwala, T. Water Demand Forecasting Using Multi-Layer Perceptron and Radial Basis Functions. In Proceedings of the 2007 International Joint Conference on Neural Networks, Orlando, FL, USA, 12–17 August 2007; pp. 13–18. [ Google Scholar ]
• Altunkaynak, A.; Nigussie, T.A. Monthly Water Consumption Prediction Using Season Algorithm and Wavelet Transform–Based Models. J. Water Resour. Plan. Manag. 2017 , 143 , 04017011. [ Google Scholar ] [ CrossRef ]
• Ahmadi, F.; Ghasemlounia, R.; Gharehbaghi, A. Machine Learning Approaches Coupled with Variational Mode Decomposition: A Novel Method for Forecasting Monthly Reservoir Inflows. Earth Sci. Inform. 2024 , 17 , 745–760. [ Google Scholar ] [ CrossRef ]
• Ebrahim Banihabib, M.; Mousavi-Mirkalaei, P. Extended Linear and Non-Linear Auto-Regressive Models for Forecasting the Urban Water Consumption of a Fast-Growing City in an Arid Region. Sustain. Cities Soc. 2019 , 48 , 101585. [ Google Scholar ] [ CrossRef ]
• Kalra, A.; Miller, W.P.; Lamb, K.W.; Ahmad, S.; Piechota, T. Using Large-Scale Climatic Patterns for Improving Long Lead Time Streamflow Forecasts for Gunnison and San Juan River Basins. Hydrol. Process. 2013 , 27 , 1543–1559. [ Google Scholar ] [ CrossRef ]
• Weekaew, J.; Ditthakit, P.; Kittiphattanabawon, N. Reservoir Inflow Time Series Forecasting Using Regression Model with Climate Indices. In Recent Advances in Information and Communication Technology 2021 ; Meesad, P., Sodsee, D.S., Jitsakul, W., Tangwannawit, S., Eds.; Springer International Publishing: Cham, Switzerland, 2021; pp. 127–136. [ Google Scholar ]
• Maddu, R.; Pradhan, I.; Ahmadisharaf, E.; Singh, S.K.; Shaik, R. Short-Range Reservoir Inflow Forecasting Using Hydrological and Large-Scale Atmospheric Circulation Information. J. Hydrol. 2022 , 612 , 128153. [ Google Scholar ] [ CrossRef ]
• Hochreiter, S.; Schmidhuber, J. Long Short-Term Memory. Neural Comput. 1997 , 9 , 1735–1780. [ Google Scholar ] [ CrossRef ]
• Kratzert, F.; Klotz, D.; Brenner, C.; Schulz, K.; Herrnegger, M. Rainfall–Runoff Modelling Using Long Short-Term Memory (LSTM) Networks. Hydrol. Earth Syst. Sci. 2018 , 22 , 6005–6022. [ Google Scholar ] [ CrossRef ]
• Filho, J.V.; Scortegagna, A.; de Sousa Dias Vieira, A.P.; Jaskowiak, P.A. Machine Learning for Water Demand Forecasting: Case Study in a Brazilian Coastal City. Water Pract. Technol. 2024 , 19 , 1586–1602. [ Google Scholar ] [ CrossRef ]
• Authors: Guo, G. Short-Term Water Demand Forecast Based on Deep Neural Network: (029). In Proceedings of the CCWI Joint Conference 2018, Kingston, ON, Canada, 23–25 July 2018; Volume 1. [ Google Scholar ]
• Le, X.-H.; Ho, H.V.; Lee, G.; Jung, S. Application of Long Short-Term Memory (LSTM) Neural Network for Flood Forecasting. Water 2019 , 11 , 1387. [ Google Scholar ] [ CrossRef ]
• Arsenault, R.; Martel, J.-L.; Brunet, F.; Brissette, F.; Mai, J. Continuous Streamflow Prediction in Ungauged Basins: Long Short-Term Memory Neural Networks Clearly Outperform Traditional Hydrological Models. Hydrol. Earth Syst. Sci. 2023 , 27 , 139–157. [ Google Scholar ] [ CrossRef ]
• Wu, T.-Y.; Juang, H.-M.H.; Chen, Y.-L.; Liu, P.-Y.; Lin, S.-I.; Chen, J.-H.; Lu, M.-M. CWB CFS 1-Tier Hindcast Analysis and Forecast Verification. In Proceedings of the 43rd NOAA Annual Climate Diagnostics and Prediction Workshop, Santa Barbara, CA, USA, 23–25 October 2018. [ Google Scholar ]
• Paek, H.; Yu, J.-Y.; Hwu, J.-W.; Lu, M.-M.; Gao, T. A Source of AGCM Bias in Simulating the Western Pacific Subtropical High: Different Sensitivities to the Two Types of ENSO. Mon. Weather. Rev. 2015 , 143 , 2348–2362. [ Google Scholar ] [ CrossRef ]
• Saha, S.; Nadiga, S.; Thiaw, C.; Wang, J.; Wang, W.; Zhang, Q.; den Dool, H.M.V.; Pan, H.-L.; Moorthi, S.; Behringer, D.; et al. The NCEP Climate Forecast System. J. Clim. 2006 , 19 , 3483–3517. [ Google Scholar ] [ CrossRef ]
• Raje, D.; Mujumdar, P. A Comparison of Three Methods for Downscaling Daily Precipitation in the Punjab Region. Hydrol. Process. 2011 , 25 , 3575–3589. [ Google Scholar ] [ CrossRef ]
• Wu, W.; Liu, Y.; Ge, M.; Rostkier-Edelstein, D.; Descombes, G.; Kunin, P.; Warner, T.; Swerdlin, S.; Givati, A.; Hopson, T.; et al. Statistical Downscaling of Climate Forecast System Seasonal Predictions for the Southeastern Mediterranean. Atmos. Res. 2012 , 118 , 346–356. [ Google Scholar ] [ CrossRef ]
• Hosseini, R.; Golian, S.; Yazdi, J. Evaluation of Data-Driven Models to Downscale Rainfall Parameters from Global Climate Models Outputs: The Case Study of Latyan Watershed. J. Water Clim. Chang. 2018 , 11 , 200–216. [ Google Scholar ] [ CrossRef ]

Click here to enlarge figure

Share and Cite

Lin, H.-Y.; Lee, S.-H.; Wang, J.-H.; Chang, M.-J. Utilizing Artificial Intelligence Techniques for a Long–Term Water Resource Assessment in the ShihMen Reservoir for Water Resource Allocation. Water 2024 , 16 , 2346. https://doi.org/10.3390/w16162346

Lin H-Y, Lee S-H, Wang J-H, Chang M-J. Utilizing Artificial Intelligence Techniques for a Long–Term Water Resource Assessment in the ShihMen Reservoir for Water Resource Allocation. Water . 2024; 16(16):2346. https://doi.org/10.3390/w16162346

Lin, Hsuan-Yu, Shao-Huang Lee, Jhih-Huang Wang, and Ming-Jui Chang. 2024. "Utilizing Artificial Intelligence Techniques for a Long–Term Water Resource Assessment in the ShihMen Reservoir for Water Resource Allocation" Water 16, no. 16: 2346. https://doi.org/10.3390/w16162346

Article Metrics

Article access statistics, further information, mdpi initiatives, follow mdpi.

Subscribe to receive issue release notifications and newsletters from MDPI journals

How to integrate artificial intelligence into office software: the ONLYOFFICE Docs case study

Share this article

AI and office software: best integration examples

Onlyoffice docs and ai: a brief overview, structure of the chatgpt plugin.

With artificial intelligence taking deep roots around our lives and making it easier for us to do a lot of things, more and more software developers, IT companies and start-ups are looking for ways to integrate state-of-the-art AI technology into their products to get the jump over their competitors.

Many modern CRM platforms, email clients, personal assistants, video editors, project management programs and other kinds of software tools now come equipped with AI assistants allowing their users to work faster and be more productive.

Office software is not an exception. The idea of integrating artificial intelligence into electronic document workflows has been winning the minds of office software developers over the last few years and now there are several interesting integration examples that significantly change the way people work with office files.

In this article, we will take a look at the most successful examples of symbiosis between artificial intelligence and office software, and examine the process of integrating an AI assistant into an office package through the example of ONLYOFFICE Docs, an open-source office suite.

The market of office software is highly competitive, and it’s an evident fact that the biggest corporations always have an advantage over other players. However, even small companies and independent developers can come up with elegant solutions bringing the power of artificial intelligence into the universe of document editing and collaboration. 

Here is a quick overview of some of the most popular office packages with integrated AI assistants available on the market:

• Microsoft 365 . The Redmond giant presented its own AI tool known as Copilot in March 2023. It combines the power of large language models (LLMs) and makes it easier for users to work with the Microsoft 365 apps, including Teams, Outlook and Edge. Copilot is deeply integrated into the Microsoft ecosystem and allows you to write and edit text content in Word, analyze trends and create comprehensive data visualizations in Excel, and create impressive presentations based on your prompts in PowerPoint. Copilot for Microsoft 365 is available with a paid subscription.
• WPS Office . Microsoft Office’s competitor from China, WPS Office, has an AI assistant called WPS AI. It’s a generative AI tool that allows users to generate text content, summarize large documents, analyze data, extract outlines from PDFs and create meeting drafts. WPS AI can be used in text documents and PDF files and offers a free trial with a limited set of features. To access its full functionality, you need to purchase one of the paid versions.
• Google Workspace . The Google corporation has an AI-powered assistant for its productivity apps called Gemini. This tool is built into Docs, Sheets and Slides allowing you to write various kinds of content and generate images in text documents, create tables and formulas in spreadsheets, and generate new slides with images and content in presentations. Gemini for Google Workspace is available as a paid add-on. Interestingly, the Google Workspace Marketplace also offers different AI assistants developed by third-party developers for Docs, Sheets and Slides but you should try them at your own risk.
• Zoho WorkDrive . This online content collaboration platform and storage has an IA integration feature called Zia. It’s an AI-based writing assistant available in some Zoho products, including Zoho Writer. Using Zia, you can check your content for spelling and grammar mistakes and improve its readability. The main objective of this tool is to help you write concisely and clearly. Zia also can generate new content and images. It’s not available for free.

Taking everything into consideration, the following conclusion seems evident: the most popular office suites provide excellent AI integration options but almost all of them are paid.

Now let’s explore the case of ONLYOFFICE Docs, an open-source office package, that provides robust AI capabilities based on ChatGPT and discover how this integration works.

ONLYOFFICE Docs is an open-source and free office suite for text documents, spreadsheets, presentations, fillable forms and PDF files. The suite has a self-hosted version for local deployment and a cloud-based version for a quick start. There is also a desktop client for Linux, Windows and macOS and mobile apps for Android and iOS. The source code of the ONLYOFFICE suite is available on GitHub .

ONLYOFFICE Docs has an open API, which makes it possible to integrate the editors with third-party services. Such integrations work via plugins, special add-ons that bring new capabilities and features. Among dozens of ready-to-use plugins for the ONLYOFFICE suite, you can find those that enable the power of artificial intelligence. More precisely, these are the plugins for ChatGPT and Zhipu Copilot. 

Both ChatGPT and Zhipu Copilot are accessible in the ONLYOFFICE editor’s interface via separate plugins that can be installed and deleted with a few clicks via the Plugin Manager. These plugins are officially developed and maintained by the ONLYOFFICE team. As all other plugins, they are available for free.

To make these plugins work, you need to specify an API key provided by the corresponding platform. When it comes to ChatGPT, you can find a valid API key in the settings of your OpenAI account. 

When you enter a valid API key in the ChatGPT plugin and enable it via the Plugins tab, you will have access to the following features via the context menu, which makes it easier to work with texts*:

• Generation of text content based on your prompts;
• Creation of images in different resolutions;
• Translation to several languages, including English, Spanish, Italian, Chinese, German, French, etc.;
• Word analysis;
• Extraction of keywords;
• Text summarization;
• Text manipulation;
• Correction of spelling and grammar mistakes;
• Communication with the chatbot.

The Zhipu Copilot plugin provides similar writing assistance features in ONLYOFFICE Docs and is designed for Chinese-speaking users because it’s based on a localized knowledge base.

In ONLYOFFICE Docs, you can interact with the ChatGPT service not only in text documents but also in spreadsheets and presentations, so you can perform various tasks such as data analysis and finding information for your slides.

Now that you know what ChatGPT and Zhipu copilot can do for you when you work on office files, let’s take a deeper look at the ChatGPT plugin to see how it works and what key elements its source code includes.

Note: the ONLYOFFICE developers assume no responsibility for the accuracy or reliability of the information provided by ChatGPT and ZhiPu Copilot.

The ChatGPT plugin consists of five directories, with the HTML files stored separately in the root directory. Here is a quick overview of each file with code samples:

1. index.html : Since the plugin operates as a background plugin, it doesn’t require a user interface. The index.html file references all the scripts, code files, and stylesheets, ensuring they are utilized when the plugin is activated.

Code inside the <head> tag of index.html file

2. chat.html : This file defines the HTML structure for the chatbox that appears when you select the chat option in the ChatGPT plugin.

3. Other HTML files : The root directory contains additional HTML files for various error messages and logs. These files are designed to handle conditions such as insufficient tokens or invalid requests, to make sure the plugin responds appropriately to different scenarios.

Overview of the resources directory

The resources directory primarily comprises two components: CSS styles and images for different modes.

1. CSS Sub-directory : This contains styles.css and custom.css . Both stylesheets apply CSS to various components throughout the plugin.

2. Image Directories : The second component includes the following:

• light contains images optimized for light-themed editors;
• dark contains images optimized for dark-themed editors;
• icons includes icons in various aspect ratios and sizes for different display settings and themes.

These above-mentioned resources ensure that the ChatGPT plugin functions well across different editor themes and resolutions.

Overview of the scripts directory

The code.js file contains the core logic for the plugin that integrates various functionalities powered by the OpenAI API. Let’s go through this file in more detail:

1. Initialization and Setup :

• The plugin initializes variables, including an API key, model settings, and UI elements.
• The checkApiKey method retrieves the OpenAI API key from local storage.

A code snippet from checkApiKey() method

2. Context Menu Generation :

• The getContextMenuItems() method dynamically generates context menu items based on the type of selection in the document (e.g., text selection, images, shapes, hyperlinks, etc.).
• If the API key is present, additional options like spelling and grammar correction, text rewriting, translation, and image generation are added to the context menu of the plugin.
• This method also includes a logic to check for the presence of images in the document and an option for generating image variations.

A code snippet from getContextMenuItems() method

3. Event Handling :

• The plugin attaches event handlers for context menu actions. For example, the onContextMenuShow() event triggers the display of the custom context menu and performs additional checks, such as retrieving the current word.
• The onSettings() event handler opens a settings modal dialogue when the settings menu item is clicked, and more.

4. Utility Methods :

• Helper methods like generateText() for text translation, showError() for error display, and isEmpyText() for text validation are included in the code.js file as well.
• The createSettings() method creates a UI container displaying token information and a settings button.

A code snippet from createSettings() method

The next file is chat.js . It manages the logic for the chatbox, which can be initialized by right-clicking anywhere in the document. It works in combination with the chat.html file to ensure the chatbox’s structure and functionality are implemented correctly.

Code snippet of the createMessage() method from the chat.js file

Now a few words about the settings.js file. It manages the settings section of the plugin. This is where users enter their API keys. This file also validates the API key entered by the user.

Code snippets of createError() and createLoader() methods from the settings.js file

There are also some other .js files in the scripts directory that mainly include JavaScript for error prompts (as mentioned earlier in the HTML section). Additionally, some files contain JavaScript triggers for different environments where the plugin might be used (desktop version, cloud version, etc.).

Translations directory

This directory contains translation files for different languages, each represented by a .json file. Using the onTranslate() method, the plugin checks this directory, identifies the file corresponding to the system’s language and retrieves the necessary translated text.

This mechanism ensures that the plugin can dynamically adapt to various languages, providing a localized and user-friendly experience.

The vendor and the licenses directory

The vendor directory houses the code and resource files for the third-party libraries used in the plugin. This directory ensures that all external dependencies are neatly organized and easily accessible. 

This directory has three components:

1. OpenAI (Chat GPT BPE Encoder Scripts) : these scripts are essential for encoding and decoding the text returned by the GPT engine.

2. Select2 Library Scripts : this is a powerful library that enhances the plugin’s user interface by providing customizable select boxes, making the plugin easier to use.

3. jQuery Base File : jQuery simplifies HTML document traversal and manipulation, event handling, and animation, making the UI more dynamic and responsive.

This was a detailed overview of the ChatGPT plugin created by the ONLYOFFICE developers for their office suite. If you want to explore the plugin’s code in detail and the methods it uses, you can take a loot at this GitHub page .

Using the same principles and the ONLYOFFICE API , you can build a plugin for any other AI-based writing assistant and use its capabilities within the interface of an office suite. 

Severe deviation in protein fold prediction by advanced AI: a case study

• Find this author on Google Scholar
• Find this author on PubMed
• Search for this author on this site
• ORCID record for Jacinto Lopez-Sagaseta
• For correspondence: [email protected]
• Info/History
• Preview PDF

Artificial intelligence (AI) and deep learning are making groundbreaking strides in protein structure prediction. AlphaFold is remarkable in this arena for its outstanding accuracy in modelling proteins fold based solely on their amino acid sequences. In spite of these significant advances, experimental structure determination remains critical. Here we report severe deviations (>30 angstroms) between the experimental structure of a two-domain protein and its equivalent AI-prediction. Severe divergence between experimental structures and AI-predicted models echoes the presence of unusual conformations, insufficient training data and high complexity in protein folding that can ultimately lead to current limitations in protein structure prediction.

Competing Interest Statement

The authors have declared no competing interest.

https://www.rcsb.org/structure/8OVQ

View the discussion thread.

Thank you for your interest in spreading the word about bioRxiv.

NOTE: Your email address is requested solely to identify you as the sender of this article.

Citation Manager Formats

• EndNote (tagged)
• EndNote 8 (xml)
• RefWorks Tagged
• Ref Manager
• Tweet Widget
• Facebook Like
• Google Plus One

Subject Area

• Biochemistry
• Animal Behavior and Cognition (5539)
• Biochemistry (12595)
• Bioengineering (9478)
• Bioinformatics (30899)
• Biophysics (15886)
• Cancer Biology (12965)
• Cell Biology (18559)
• Clinical Trials (138)
• Developmental Biology (10026)
• Ecology (15008)
• Epidemiology (2067)
• Evolutionary Biology (19197)
• Genetics (12768)
• Genomics (17573)
• Immunology (12719)
• Microbiology (29774)
• Molecular Biology (12400)
• Neuroscience (64892)
• Paleontology (480)
• Pathology (2008)
• Pharmacology and Toxicology (3468)
• Physiology (5355)
• Plant Biology (11125)
• Scientific Communication and Education (1729)
• Synthetic Biology (3065)
• Systems Biology (7702)
• Zoology (1732)

• Application Requirements
• Student Experience Q&A
• Business Analytics Certificate
• Finance Certificate Programs
• Tuition & Financial Aid
• International and Transfer Students
• Military Friendly
• How to Apply
• Accreditation
• Leadership Development Resources
• MBA Career Outcomes
• MSBA Career Outcomes

A Guide to Artificial Intelligence in the Enterprise

The explosion of artificial intelligence has been extraordinary. Over the past few years, new tools such as ChatGPT, DALL-E 2, and Gemini opened the floodgates of generative AI. They make tasks like drafting an email or creating an image extremely simple. However, AI is much more than a writing or image-generation shortcut. The real benefactors of AI are enterprises. Since AI can analyze vast amounts of data and draw conclusions, it may change how we work across all areas of business.

This post explores AI enterprise use cases, including how to deploy AI in the enterprise and potential challenges you may face.

Understanding Artificial Intelligence

First, what is enterprise AI?

Artificial intelligence refers to computer systems that are capable of performing work that’s usually handled by humans, such as identifying patterns, solving problems, recognizing speech, or making decisions. AI uses several techniques, including deep learning, machine learning, and natural language processing (NLP), to complete and resolve tasks. 1

Organizations that incorporate AI into their operations use various forms of publicly available or in-house AI enterprise software. The tools assist with specific tasks, such as automating repetitive tasks, analyzing financial data, making product recommendations or performing basic computer coding. Implementing enterprise AI tools may help organizations save time and money and reduce errors. 1

Applications of AI in the Enterprise

How will artificial intelligence for enterprise applications change how we work? Let’s look at a few use cases.

AI in Customer Service and Support

Customer service is one area where AI can make big waves. By implementing AI-powered chatbots, organizations can improve their customer engagement. Rather than contacting a human customer service agent, clients can interact with chatbots 24/7, without waiting to speak to an associate. The chatbots can resolve simple customer inquiries, and if someone does need additional assistance, it can route their question to the appropriate associate. 2

AI for Marketing and Sales Optimization

There are numerous ways to use artificial intelligence for enterprise applications in marketing. You can use it to design marketing emails and social media posts, create subject lines, and write website copy. It’s also helpful for segmenting your audience and optimizing your blog posts to boost your rankings on the major search engines. 3

AI in Supply Chains and Logistics  

Enterprise AI platforms can help with supply chain management through inventory planning, production and logistics. Such platforms can analyze massive amounts of data, helping organizations predict trends and optimize their supply chains to meet the anticipated demand. That means less unused inventory taking up warehouse space and reduced spending on manufacturing excess products. 4

AI for Financial Analysis and Risk Management

AI can handle many financial planning and analysis tasks, including ratio and variance analysis and basic reporting. Its capabilities allow it to review vast amounts of data and identify anomalies and patterns. It’s also useful for forecasting revenue, demand, expenses, and other critical financial details. 5

AI in Human Resources and Talent Management

AI can help with many HR tasks, such as writing job descriptions, simplifying the onboarding process, and assisting in workforce planning. 6

AI for IT and Cybersecurity

AI for enterprises can assist with software quality assurance (QA) testing. It may also aid in identifying potential cybersecurity threats and writing code for software and websites. 7,8

AI in Product Development and Innovation

Using enterprise AI, a company can gather market trend data to explore opportunities for new products. It can also assist in refining product designs and testing how a product works under different scenarios. 9

Implementing AI at Your Organization

Clearly, there are many use cases for AI in enterprises. Business managers can harness the technology by understanding its capabilities and identifying areas where it may be beneficial from a cost and productivity perspective.

Managers who are unfamiliar with AI may be uncomfortable incorporating it into their daily tasks. However, most AI tools don’t have quite the learning curve that some earlier technologies presented. The best way to test it is to experiment with the available tools to see how they fit into your workflow. A more advanced implementation of AI across entire departments will require additional in-depth testing and perhaps in-house development. 10

Building an AI-Ready Culture

In the past, significant swathes of the workforce didn’t need to understand advanced IT concepts like coding or data analysis. That will likely change as more organizations deploy AI for enterprise applications.

Some AI platforms may require knowledge of business intelligence tools, like Tableau, or basic programming knowledge using a language like Python. Businesses can prepare their teams for the future by upskilling and reskilling employees on the latest data analysis best practices. They can also encourage their employees to learn how to use efficient prompts with generative AI. 11

Challenges and Considerations

Integrating AI presents several challenges. One is a skills gap. As a new technology, many workers are unfamiliar with AI’s capabilities and how it can improve the business’s processes. Companies must encourage their employees to boost their AI skills through training and development. Another issue is identifying where AI can add the most business value. Focusing on specific problems can help a company avoid unnecessary spending on AI tools that don’t yield a positive return on investment. 12

Case Studies of Successful Enterprise AI Applications

AI can bring significant benefits, including improved efficiency, customer satisfaction and cost savings. The following case studies illustrate that successful AI deployment in enterprises requires careful planning, effective data management and addressing the specific challenges of each use case.

Case Study 1: Predictive Maintenance in Manufacturing

Siemens implemented AI-driven predictive maintenance in their manufacturing processes to reduce downtime and increase efficiency. By using sensors and advanced analytics, the AI system can predict equipment failures before they happen. 13

Deployment Strategy:

• Data Collection: Sensors were installed on machinery to collect operational data in real time
• AI Model Development: Siemens developed machine learning models that analyze historical and real-time data to predict failures
• Integration: The AI solution was integrated with existing maintenance management systems to automate scheduling and alerts for upkeep activities

Challenges:

• Data Quality: Ensuring the data collected from sensors was accurate and relevant
• Change Management: Training employees to trust and use the new AI-driven maintenance schedules
• Scalability: Ensuring the AI models could be scaled across different plants and machines

Outcomes: 14

• Reduced Downtime: Machine downtime was reduced by up to 50%, resulting in significant cost savings
• Increased Lifespan: The lifespan of machinery increased due to timely maintenance
• Enhanced Efficiency: The overall operational efficiency saw a noticeable uptick

Case Study 2: Customer Service Automation in Retail

H&M deployed AI to enhance its customer service by implementing chatbots and virtual assistants that handle customer inquiries across multiple channels, including their website and social media. 15

• AI Training: Chatbots were trained using historical customer service data to understand common queries and appropriate responses
• Pilot Testing: The AI solutions were first tested in low-risk scenarios to gauge performance
• Full Implementation: Following the successful pilot, the chatbots were rolled out across multiple customer touchpoints
• Language and Context: Ensuring the AI could understand and respond accurately in various languages and contexts
• Customer Acceptance: Managing customer expectations and promoting the use of AI chatbots
• Continual Learning: Incorporating a feedback loop for continuous improvement of the AI responses
• Improved Response Time: Customer inquiries were addressed faster, cutting wait times by 70% and enhancing customer satisfaction 15
• Cost Savings: Reduced the need for human agents, enabling the reallocation of resources to more complex tasks
• Scalable Support: The AI system can handle an exponentially larger number of inquiries during busy periods compared to human staff

Case Study 3: Personalized Marketing in Financial Services

Capital One used AI to provide personalized marketing and customer engagement strategies. They could deliver highly targeted marketing messages and offers by analyzing customer transaction data. 16

• Data Analysis: Leveraged big data and machine learning to analyze customer behavior and transaction histories
• Segmentation: Customers were segmented based on patterns identified by the AI models
• Personalized Outreach: Personalized marketing campaigns were developed for each segment
• Privacy Concerns: Ensuring customer data is handled responsibly and complies with regulations
• Algorithm Bias: Avoiding biases in the AI models to ensure fair and equitable marketing practices
• Integration: Seamlessly integrating AI insights with existing CRM systems
• Increased Engagement: Higher engagement rates for marketing campaigns due to personalization
• Enhanced Customer Experience: Customers received offers and communications that were relevant to their needs
• Revenue Growth: Improved campaign effectiveness translated to a measurable increase in revenue

Measuring AI’s Impact and Performance

As companies implement AI enterprise tools, monitoring their effectiveness is critical. Start by defining several key performance indicators (KPIs) for each task involving AI. The metrics will vary by use case but may include evaluation for accuracy, productivity improvements, and cost savings. You can also request feedback from employees to learn their thoughts about the tools. Such input can help you refine your existing business processes and find new ways to implement the technology. 17

As businesses deploy new AI technologies, it’s more important than ever for managers to understand how AI can add value. Obtaining an Online MBA from William & Mary can help you acquire the skills to evaluate these enterprise AI opportunities. Learn on your own schedule from anywhere you have an internet connection. Our expert faculty will help you grow into a skilled business leader while you develop connections that will last your entire career.

To learn more, schedule a call with an admissions outreach advisor.

• Retrieved on August 7, 2024, from  coursera.org/articles/what-is-artificial-intelligence
• Retrieved on August 7, 2024, from  upwork.com/resources/how-is-ai-used-in-business
• Retrieved on August 7, 2024, from  ibm.com/blog/ai-in-marketing/
• Retrieved on August 7, 2024, from  ibm.com/think/topics/ai-supply-chain
• Retrieved on August 7, 2024, from  corporatefinanceinstitute.com/resources/fpa/ai-for-financial-analysis/
• Retrieved on August 7, 2024, from  resources.workable.com/tutorial/artificial-intelligence-in-human-resources
• Retrieved on August 7, 2024, from  deloitte.com/us/en/insights/focus/tech-trends/2022/future-of-cybersecurity-and-ai.html
• Retrieved on August 7, 2024, from  softengi.com/blog/ai-in-it-how-artificial-intelligence-will-transform-the-it-industry/
• Retrieved on August 7, 2024, from  dovetail.com/product-development/ai-in-product-development/
• Retrieved on August 7, 2024, from  hbr.org/2023/07/build-a-winning-ai-strategy-for-your-business
• Retrieved on August 7, 2024, from  deloitte.com/us/en/pages/technology/articles/build-ai-ready-culture.html
• Retrieved on August 7, 2024, from  forbes.com/sites/forbesbusinesscouncil/2023/10/24/11-challenges-of-adopting-ai-in-business-and-how-to-address-them-head-on/
• Retrieved on August 19, 2024, from siemens.com/global/en/products/services/digital-enterprise-services/analytics-artificial-intelligence-services/predictive-services.html   
• Retrieved on August 19, 2024, from blog.siemens.com/2023/07/predictive-maintenance-at-scale-is-entering-the-mainstream/  
• Retrieved on August 19, 2024, from tidio.com/blog/companies-using-ai-for-customer-service/  
• Retrieved on August 19, 2024, from capitalone.com/tech/machine-learning/  
• Retrieved on August 7, 2024, from  impactmind.ai/blog/measuring-the-success-of-your-ai-implementation-metrics-and-best-practices

Return to Online Business Blog

William & Mary has engaged Everspring , a leading provider of education and technology services, to support select aspects of program delivery.

Online assessment in the age of artificial intelligence

• Open access
• Published: 19 August 2024
• Volume 3 , article number  126 , ( 2024 )

Cite this article

You have full access to this open access article

• Alexander Stanoyevitch 1  

12 Accesses

Explore all metrics

Online education, while not a new phenomenon, underwent a monumental shift during the COVID-19 pandemic, pushing educators and students alike into the uncharted waters of full-time digital learning. With this shift came renewed concerns about the integrity of online assessments. Amidst a landscape rapidly being reshaped by online exam/homework assistance platforms, which witnessed soaring stocks as students availed its questionable exam assistance, and the emergence of sophisticated artificial intelligence tools like ChatGPT, the traditional methods of assessment faced unprecedented challenges. This paper presents the results of an observational study, using data from an introductory statistics course taught every semester by the author, and delves into the proliferation of cheating methods. Analyzing exam score results from the pre and post introduction of ChatGPT periods, the research unpacks the extent of cheating and provides strategies to counteract this trend. The findings starkly illustrate significant increases in exam scores from when exams of similar difficulty were administered in person (pre-Covid) versus online. The format, difficulty, and grading of the exams was the same throughout. Although randomized controlled experiments are generally more effective than observational studies, we will indicate when we present the data why experiments would not be feasible for this research. In addition to presenting experimental findings, the paper offers some insights, based on the author's extensive experience, to guide educators in crafting more secure online assessments in this new era, both for courses at the introductory level and more advances courses The results and findings are relevant to introductory courses that can use multiple choice exams in any subject but the recommendations for upper-level courses will be relevant primarily to STEM subjects. The research underscores the pressing need for reinventing assessment techniques to uphold the sanctity of online education.

Similar content being viewed by others

The quizzical failure of a nudge on academic integrity education: a randomized controlled trial

Guidelines for Creating Online MCQ-Based Exams to Evaluate Higher Order Learning and Reduce Academic Misconduct

Explore related subjects.

• Artificial Intelligence

Avoid common mistakes on your manuscript.

1 Introduction

Distance learning dates back to 1728, when Caleb Phillips advertised in the Boston Globe newspaper his shorthand course offering by mail correspondence for students throughout the United States; see [ 1 ]. The internet provided distance learning with a tremendous boost by greatly reducing the turn-around times between professor-student correspondences. The development of new technologies has significantly increased the quality of online courses over the years since. The single most significant catalyst for online education came with the pandemic, where every instructor, from primary education through graduate school levels, needed to quickly develop sufficient expertise in order to offer online versions of all of their courses that were forced upon them.

The pandemic coincided with, perhaps, the most profound revolution in distance learning ever. Students and instructors who never had any experience with online learning suddenly became full-time users. Debates flourished and studies came out in droves of the effectiveness of online courses. A nice study of the impact of Covid-19 on 309 courses given at eight colleges and universities spanning four continents is given in Bartolic et al. [ 2 ].

A crucial concern regarding online courses is the integrity of assessments. Before the pandemic, many instructors resisted the idea of offering their courses online, fearing that assessments would be compromised. The reality is that, compared to administering exams in a classroom setting under the watchful eye of a proctor, allowing students to take exams online at home makes it almost impossible to prevent cheating. There are evident actions students might take during an online exam, with or without permission: opening their books and notes, consulting the web, collaborating with classmates, or seeking assistance from advanced students. Several companies, who ostensibly advertise that they offer help with homework, had no qualms about helping students cheat on exams. During the pandemic, several of these companies and their shareholders profited immensely from this dubious practice; One of them saw their market capitalization nearly quadrupling from the onset of the pandemic to its zenith. For a modest monthly fee, students can obtain one-on-one assistance from a tutor, typically lowly paid from a third-world country, to complete their exams and gain access to an extensive database of solved exam problems across various subjects. A colleague of the author purchased such a subscription from one of these companies to understand the full extent it could undermine online assessment, and the findings were alarming. This instructor discovered that their exam questions, complete with answers, appeared on the platform the very next day!

The focus of this paper concerns assessing student performance in online courses when the exams are administered without any sort of online proctoring. In a face-to-face class, when we proctor an exam in the classroom, we have complete control over what resources students can use. For example, for a math exam, we typically will allow calculators, but neither laptops nor cellphones. If the same exam is given online without supervision, there are many things that cannot be controlled. Despite instructions, students might use their cellphones or computers to access online resources or even network with classmates on the exam questions. Our paper will provide compelling evidence of the prevalence of cheating with online exams using data from the author’s introductory large lecture statistics courses, which he has been teaching every fall and spring semesters for over a decade, with 120–150 students each semester.

Another stumbling block in ensuring the integrity of online assessments is the recent emergence of artificial intelligence-driven large language models, most notably ChatGPT, which is owned and hosted by Microsoft. Google has developed its own large language model, known as Gemini. Unlike services that employ tutors and maintain a database of exam questions and solutions, AI-based large language models have been trained on vast data sets and can answer questions from a myriad of contexts with remarkable precision. ChatGPT offers free access to an earlier version (Chat GPT 3.5) with a limit on the number of questions within a certain timeframe. Subscriptions are available that eliminate these restrictions and use a more powerful model (Chat GPT 4).

Figure 1 , sourced from OpenAI's paper introducing their ChatGPT version 4, illustrates its performance across a range of well-known exams:

This figure is taken from the GPT-4 Technical Report by OpenAI (the company behind ChatGPT) accompanying the introduction of ChatGPT-4 in February 2023. The performance percentiles on various tests are depicted: the blue bars represent the previous version, GPT-3.5, and the green bars represent GPT-4. For instance, on the GRE-Verbal exam (a national test used for graduate school admissions), GPT-3.5 scored better than approximately 60% of the students taking the exam, whereas GPT-4 outperformed about 98%

This graphic shows that most all standardized exams cannot be administered without proper proctoring.

In this paper, we explore the present landscape and the future of assessment in online courses, considering third party online homework/exam assistance platforms and ChatGPT. Often, it is neither practical nor possible to proctor exams in an online format. Thus, this issue holds significant importance for the future of online education. Our purpose here will be to provide guidance and advice for giving online proctored exams in the era of large language models such as Chat GPT. Most of this advice stems from the author’s extensive experience and findings in using online assessments over the past four years, in mathematics and statistics courses, both at the introductory and advanced undergraduate levels. The key takeaway is that while it remains feasible to conduct online assessments, ensuring the integrity of exams requires a substantial amount of additional effort. This state of affairs will certainly impact university and individual instructor decisions on whether to use online assessment.

2 Review of related literature

For many years before the pandemic, internet-based distance learning has been offered and developed at many universities throughout the world. It has made education more equitable by allowing students to take college classes (and earn degrees) without having to make a major and economically difficult move, waste time with long commutes, and also to be able to schedule classes around their jobs or family obligations. According to the Oxford Learning College [ 1 ], online courses date back to 1965 (before the internet) when the University of Alberta used networked IBM 1500 computers. Also, Massive Open Online Courses (MOOCs) were first offered through MIT in 2012. In 2019, the last year before the pandemic hit, the number of higher education students taking online courses was 36.9%; see [ 3 ]. With the increasing prevalence of online education, much research has been done over the years. The meta-study by Bernard et al. [ 4 ] examined 232 separate studies of attitudes, achievement and retention outcomes and contains numerous useful research paper references. Horspool and Lange [ 5 ] compared online versus face-to-face learning and examined student’s perceptions, behaviors, and success rates. Their findings indicate that schedule flexibility is a major factor when students choose an online course over a face-to-face option and that students in both formats felt that they experience high-quality communications with the instructor. They also found that online students study more at home than face-to-face students, but had more limited peer-to-peer communication. The authors present several recommendations, based on their findings, to improve the online course experience. The Horizon 2020 TeSLA Project [ 6 ], funded by the European Union, is a valuable resource for online learning. In particular, the Publications from this project contain a wealth of literature on the assessment topics relating to the topics of this paper. In an effort to assist institutions who are hesitant to adopt online learning course offerings, Fidalgo et al. [ 7 ] conducted an extensive survey on undergraduate students in three countries to find out their feelings and concerns about online courses (the survey was done before Covid-19). Shortly into the pandemic, Gammage et al. [ 8 ] reviewed assessment security practices in safeguarding academic integrity at several different universities.

Before the onset of COVID-19, the prevalence of online course offerings was shaped by student demand and the willingness of faculty and universities to provide this instructional mode. Online assessment has consistently posed challenges, with difficulties in ensuring assessment integrity discouraging many faculty members from embracing online education. With COVID-19's arrival, faculty found themselves unprepared in the midst of the spring semester (or quarter) of 2020. Traditional face-to-face instruction came to a sudden halt, compelling all instructors to rapidly transition to online teaching. Unfortunately, the outcomes were not always favorable.

Determining the exact extent to which students cheat or even anticipating all possible methods they might employ is challenging. Yet, studies indicate a notable surge in the number of students admitting to cheating on online exams during the COVID-19 pandemic. A recent meta-study by Newton and Essex [ 9 ], encompassing 19 studies and 4,672 participants from as far back as 2012, revealed that self-reported online exam cheating rose from a pre-COVID rate of 29.9 to 54.7% during the pandemic. The primary reason students cited for cheating was the mere availability of an opportunity. It's pivotal to recognize that such studies, dependent on voluntary responses, can introduce bias, potentially underestimating the real figures. Nevertheless, the substantial uptick during the pandemic is of significant concern. Noorbehbahani et al. [ 10 ] carried out a meta-study reviewing 58 papers on online cheating from 2010 to 2021. Their work aimed to summarize patterns in cheating types, detection methods, and prevention strategies in online environments.

In Dendir and Maxwell [ 11 ], compared student scores on the same online exams when some were administered without proctoring and others with online proctoring. They did this comparison in two different courses: Economics and Geography, throughout the semester for a total of 3 high-stakes exams for each class. Their analysis showed significantly higher scores for the non proctored exams over the online proctored exams, over both courses and for all three exams. Many college instructors make use of test banks to create multiple choice exams. Such questions can promptly appear in databases of companies that provide assistance to students on exams and homework. Golden and Kohlbeack [ 12 ] did experiments with their online accounting exams and found that if test bank questions are modified by paraphrasing, the student scores on them dropped on average from 80 to 69%. This paper was published before the release of Chat GPT, which can serve as another powerful tool to allow students to cheat on exams.

This paper will showcase the outcomes of experiments designed to gauge student cheating, particularly with the utilization of online student test assistance platforms during the pandemic, in the context of the author's large introductory statistics lecture. The findings indicate that cheating via such platforms was rampant. Each semester, the author typically offers one upper-level course alongside the introductory class, with a smaller enrollment. The specialized nature of the upper-level course makes it harder for students to cheat using such platforms. However, the advent of ChatGPT exposes both course types to novel cheating techniques. This paper will delve into the diverse methods students resort to for cheating and the strategies to mitigate them. Although the primary focus is on mathematics and statistics courses, the insights should be invaluable to instructors across various disciplines.

Holden et al. [ 13 ] suggest distinct strategies to curb student cheating in online exams, differing from our propositions. Their expertise lies in non-STEM courses, where they recommend three types of video surveillance techniques and AI-driven plagiarism detection methods. However, video surveillance can be time-consuming, and it may pose ethical and legal challenges. Jia and He [ 14 ] at Beijing University crafted an AI-based proctoring system tailored for mathematics courses, utilizing automated video analysis through AI algorithms. They found it to be cost-efficient and highly effective in curbing student cheating in online evaluations.

A strategy closer to our paper's methodology is proposed by Nguyen et al. [ 15 ]. In the realm of Chemistry exams, they introduced several economical assessment formats that substantially reduce cheating while still meeting learning objectives.

While large language models can inadvertently assist students in cheating during online exams, they can also augment instruction in numerous beneficial ways, as demonstrated by Jeon and Lee [ 16 ].

3 Methods and results

One of the main purposes of this paper is to present evidence showing the extent of students cheating in the author’s introductory statistics class when it is offered online compared to when it is offered in the classroom. This is an observational study: the author gives this large lecture statistcs course every fall and spring semester with enrollments in the range 120–150, and has been doing so for the past 10 years. Each semester, there are four exams and a final exam, all multiple choice. Great care it taken to assure that the difficulty of each exam remains the same each semester and also the grading is done completely objectively and simply (each answer gets full credit if correct or no credit if wrong, there is no partial credit or need for any subjective assessment). We collected grade distributions for each exam over all the semesters. Prior to the pandemic, these distributions were quite similar for each exam over the different semesters, but once the pandemic hit and the exams were administered online, the scores increased markedly.

The most reliable statistical methods for collecting data are randomized controlled experiments. But such a design would not be feasible to test our hypothesis. We would need to announce during registration, that students who sign up for this class would be randomly assigned to one of two groups: one group would take their exams in class and the other online (but otherwise the classes would be identical). Few if any students would be willing to sign up for such a class and even if they did, as our research will show, the students taking the exams online would have much higher scores (on the same exams), and this would lead to chaos. Another design might be to offer two different classes: one in which the students would take their exams online and the other where they would take the exams in class. The exams and other aspects of the class would be identical. This design, however, would be flawed, since there would exist many lurking variables and differences stemming from which students choose which format. So a retrospective observational study, as we do in this paper, appears to be the best method for performing such comparisons.

As discussed above, we will present different ways that students have been able to cheat and the prevalence of such cheating using test homework help offered by several companies. We will also explore the new challenges introduced by ChatGPT.

The author has also regularly been teaching an upper-level mathematics course in machine learning every spring semester since before the pandemic. This class typically has an enrollment from 20 to 30. The exams were all take-home exams but very different in nature from the multiple choice introductory statistics exams. The questions involve mixes of mathematical and conceptual reasoning, and some involve computer programming. The recent availability of large language models like Chat GPT, has introduced some problems in assessment. Chat GPT, for example, is capable of writing computer programs. This paper will also discuss ways to prevent students from using such large-language models on exams, and these mitigation methods have been tested during several semesters.

4 Introductory large lecture statistics course

4.1 part 1a: pre-chat gpt, introductory-level large lecture statistics course.

The author’s venture into online teaching and assessment began abruptly in the middle of the spring semester of 2020 when the pandemic-induced lockdowns commenced and universities suspended all on-campus classes. All faculty at the author’s (California State University) campus were provided with some online teaching tools (including Zoom, Blackboard, and Camtasia) and were given just one week to prepare to continue all of their courses online, for an indefinite amount of time. Initially, our university anticipated this arrangement to last less than a month, an estimate that, in hindsight, proved greatly optimistic.

Prior to the pandemic, the author had essentially no experience with online education. Although there were opportunities to give online courses, the author’s reservations with online assessment along with his preference for in-person rather than online interactions deterred any further exploration of this instructional method. The swift shift to online was thrust upon us all, requiring urgent preparation and numerous decisions. The university aimed for maximum flexibility: faculty could conduct their lessons synchronously (live Zoom sessions), asynchronously (pre-recorded lectures), or a mix of both. For live sessions, instructors had the option to record and post these lessons. However, legal complications arose: was there a need to obtain student consent to upload recordings of live classes in which they actively participated? With limited time to address every issue, most began with a tentative plan and adapted as circumstances evolved. Maintaining integrity of exams and assessments was a major concern for many and for the author in particular.

For the lower-level large lecture class, the author used the same format for exams as was used in the face-to-face exams: multiple choice questions. These were now given on the Blackboard system with no proctoring rather than in a proctored classroom setting. Whereas in the in-class setting, the author did not allow the use of book, notes, or cell-phones (not to mention computers), there was really no way to prevent this in the online setting, so the author had to assume that students would be using all of these resources (despite whatever instructions were given). In addition, students could work with each other or get help from others in completing these exams. Exam collaboration with classmates can be made more difficult by scrambling both the question order as well as the order of the multiple choice answers for each individual exam —a feature available on Blackboard and other course management systems.

Blackboard, as well as CANVAS (another widely used system) has a feature called lockdown browser. This tool aims to prevent students from accessing other websites during their exams and prevents actions like copying and pasting. Although the author used this feature on his exams, there are ways around it—for example, by opening up a different browser. Using the lockdown browser also introduces potential complications. At times, it might malfunction, inadvertently preventing students from completing their exams—often through no fault of their own. While the malfunction rate isn’t alarmingly high—in the author's experience, it hovers around 2–3%—in a class of 140 students, this equates to 3–5 individuals per exam in the large lecture that require special accommodation. Moreover, the necessity for the instructor to remain available during exams to address technical glitches and also to arrange new times for students to complete or retake their exams undermines one of the intended benefits of automated online exams. Implementing this feature to reduce cheating opportunities can introduce significant collateral challenges, both in terms of time and inconvenience.

Navigating through the initial semester of Covid, while simultaneously learning the intricacies of online teaching, proved to be a substantial challenge. Most of the author’s energy was channeled into determining the most effective methods to deliver lessons and engage students. Foremost in his concerns was devising strategies to conduct exams with utmost integrity. It was evident that certain measures were essential to minimize cheating. One such measure was requiring that online exams were conducted within a narrow time slot. At first glance this may have seemed straightforward since all students should presumably be free during the designated time slot for the class. The unpredictable nature of technology meant that some students would inevitably encounter IT issues, necessitating buffer time to accommodate them. Within Blackboard, there are primarily two features to regulate the duration of an exam: one is the window during which the exam link is accessible, and the other is an inherent time limit on the exam itself.

With the first exam, I gave too much time and kept the exam difficulty level similar to that of an in-person exam. Both turned out to be major errors. The results are nicely illustrated by the following two bar plots of grade distributions for the students (in the same) class, in their first exam (that was administered in person), versus for their third exam (that was administered online); see Fig.  2 below.

NOTE: For readers outside of the United States: Grades in most US colleges and universities (as well as in primary and secondary schools) are letter grades which typically have the following meaning: A = outstanding, B = good, C = satisfactory, D = unsatisfactory, and F = failing. Grades are converted numerically with A = 4 points, B = 3 points, C = 2 points, D = 1 points, and F = 0. Using these numerical conversions, grade point averages for particular students (or groups thereof) and of exams can be computed.

Based on the author’s impression of student understanding and participation during the online lectures as well as the extremely higher grades with the online exams, this discrepancy does not seem indicative of enhanced remote learning proficiency. Although this does not present a definitive proof that students were cheating with the online exams, the author feels that the elevated grades on Exam 3 can largely be attributed to the increased opportunities for students to cheat during online assessments. There are other possible explanations, e.g., perhaps during the pandemic, students were able to study better at home with less distractions and reduced time spent on transportation. A controlled randomized experiment would reveal the true reason for the much higher grades on (the same) exams when given online, but, as we pointed out earlier, it would not be feasible to perform such an experiment.

Upon observing the results of Exam 3 during Spring 2020, it became apparent to the author that adjustments were necessary. A shorter exam duration was introduced, and the difficulty level was elevated. While these changes yielded a more “normal” grade distribution in Exam 4 (like the one on the left of Fig.  2 ), they risked unduly penalizing genuine students who refrained from any dishonest practices.

In the Fall 2020 semester some new trends became evident. Typically, the author would reuse questions on exams so as to compare student performances and gauge how certain modifications in teaching may impact student learning. The first three exams (out of a total of four plus a final exam) had similar grade distributions to the typical one (left of Fig.  2 ). This was attributed to the adjustments made by the author based on the learnings from the previous semester. This also gave a good indication that the usage of real-time tutors on exam assistance platforms was not drastically affecting the grade distribution. Perhaps most students (who used online learning platforms on exams) did not feel they needed a live tutor, relying instead the large question/answer banks that such platforms provide.

For the fourth exam, the author did a simple experiment: For Fall 2020 semester, the author gave exactly the same fourth exam as was used in a semester several years back when the course was given face-to-face. Sure enough, compared with the typical symmetric distribution (left of Fig.  2 ) with all pre-Covid exams, the grade distribution for the online exam was ridiculously skewed—A was now the most common grade, see Fig.  3 .

The left distribution had been rather typical for all exams over the many semesters that the author has taught his large lecture statistics class, following a bell-shaped distribution. The distribution on the right comes from the author’s very first online exam in Spring 2020. It has many more A’s and B’s and quite a bit less C’s and D’s

The ridiculous distribution for Exam 4, of Fall 2020, which, as an experiment, used the same questions as a previous semester’s exam a few years back. This provides compelling evidence that exam questions get added to third party exam assistance platform’s data bases of exams and that most students were using such platforms to cheat

The difference of the distribution in Fig.  3 with a typical symmetric distribution for the author’s Exam 4 pre-covid were even more extreme if we take into account the exam curves. The reason is that all exams are typically curved (with very similar curves for each exam over all pre-Covid semester). But since the numerical scores were so much higher with the online exams, their curves were less generous.

For Exam 4, which tends to be the most difficult of all of the exams (including the final exam), the scores tend to be lower and so the curve is more generous. Here are the curves for Exam 4 both before and during Covid:

Pre-Covid curve for (face-to-face) Exam 4: A: 75–100, B: 63–74, C: 45–62, D: 30–44.

Covid curve for (online) Exam 4: A: 90–100, B: 80–89, C: 70–70, D: 50–69.

This shows that the differences in grades were even more extreme than the distributions indicated. For example, pre-covid on this exam, a 75% was the bottom of the A grade range, whereas for the online version a 90% was needed for the same bottom of the A. Indeed, if the covid curve for this exam were used to assign grades to the students who took the exam face-to-face, the distribution would be skewed to the right (see the right side of Fig.  4 ), with about half of the students getting an F.

A comparison of the grade distributions for the same multiple choice exam given online and unproctored (left) versus face-to-face proctored in the classroom (right) if the same curves were used

For the online exam it would have been reasonable to make the curve even less generous: e.g. A: 95–100. But the instructor would never make exam curves stricter than the “standard curve” and before giving online exams there would have never been a need to do this, since the exams were always sufficiently challenging to necessitate some sort of a downward curve.

To counteract this, the author crafted a final exam using entirely new questions, aiming to level the playing field and diminish the advantages for those relying on external aids.

Moving on into the Spring 2021 semester, when the author taught the same large lecture statistics course online, additional experiments and modifications were introduced.

Two Experiments : The author crafted two distinct modifications for roughly half of the questions on the first exam:

The modified question retained the original wording, and the multiple-choice answers remained unchanged. However, one numerical parameter (e.g., a standard deviation) was adjusted, altering the correct answer to another existing option.

The original question's wording was preserved, but the initially correct multiple-choice answer was replaced. Along with this, a numerical parameter (e.g., a standard deviation) was adjusted to change the correct answer to a different option.

The results were quite illuminating. For Experiment (1) the average percentage of students who chose the previously correct (but now incorrect) answer for such questions was 72%! This is corroborating evidence that most students were using third party online platforms to cheat and that many did not even read the actual questions! The 72% figure probably underrepresents the number of students who cheated in this way, as some astute learners might have read the questions and cross-referenced with their online platform to validate their initial interpretations.

In Experiment (2), students solely depending on third party online platforms were likely taken aback. Some even audaciously emailed the instructor, asserting that certain multiple-choice questions lacked the correct answers (which, indeed, were present).

4.2 Part 1B: introductory-level large lecture statistics courses after chat GPT

The emergence of Chat GPT has added further challenges for educators striving to uphold the integrity of online exams. As outlined in the introduction, Chat GPT can accurately answer numerous questions when they're directly inputted into the application. Educators are advised to test some of their online or take-home exam questions on Chat GPT to gauge the accuracy of its responses.

The capability of Chat GPT as well as other evolving large language models continues to improve. Questions for which Chat GPT gives the correct answer should be changed or replaced. We will give some general suggestions, but we first present the following rather revealing example.

Original Draft Exam Question: Suppose that two dice are rolled and the numbers displayed on top are added up. The probability that this sum is at most 3 is…

The correct answer is: 1/12.

The Chat GPT answer is: 1/12 (= 3/36)—CHAT GPT Got the correct answer!!

Modified Draft Exam Question: Suppose that we have two pyramid shaped 4-sided dice, with the faces numbered 1 through 4. Suppose that the two dice are rolled and the numbers displayed on the bottom are added up. The probability that this sum is at most 3 is…

The correct answer is…3/16.

Chat GPT's Response: Incorrect.

Despite the current version of Chat GPT not answering the modified question accurately, its capability and efficiency continue to improve. It might be able to answer similar questions correctly in the future.

4.3 Timeline for the online classes above

Below is a timeline that summarizes the exam and grade information presented above:

Spring 2020 Semester (Covid-19 started): Exams switched mid-semester from face-to-face to online. Exam 3 (the first online exam) had much higher grades than usual. Adjustments were made for remaining exams to mitigate this (shorter duration, increased difficulty).

Fall 2020 Semester (first full online semester): The author continued with the above strategy as well as some new ones that were be elaborated upon in the following section (recommendations) for the first three exams and for the final exam. For Exam 4 an experiment was done. The same exam that was used several years ago was given. The grades turned out ridiculously high.

Spring 2021 (another online semester for this class) The exams continued to be modified as explained above and this kept the exam results more down to earth.

Advice for making online exams for introductory statistics and related classes : The above discussion shows the variety of online resources that students can use to cheat on unsupervised online or take-home exams.

No Repetition : Third party online exam/homework assistance platforms have demonstrated that it's imperative not to reuse questions. Even those used just a day earlier can end up on such platforms. For integrity, either:

Draft entirely new questions.

Modify previous ones by adjusting parameters and/or altering the multiple-choice options.

Tackling Live Tutoring: It's a bit more challenging to address students who employ live tutors during exams. However, imposing strict time constraints can discourage this behavior.

Chat GPT Screening: As the efficiency of tools like Chat GPT increases, it's crucial for instructors to keep abreast of such new technology and to screen exam questions using a Chat GPT session to ensure they aren’t easily answered by such platforms.

Incorporating Graphics and Unique Symbols: Questions that reference specific graphics or use specialized Greek letters in unique fonts can pose a challenge for platforms like Chat GPT. It adds an additional layer of difficulty in deciphering and interpreting the questions.

While these strategies involve more preparation, the integrity they bring to the examination process is invaluable. It's a trade-off: increased time in question creation versus the time saved from manual grading. Many educators, when given a choice, might still lean towards supervised, in-person examinations. However, for purely online courses, these precautions become essential.

5 Guidelines for online assessments in advanced mathematics and computer science courses

In general, upper-level STEM classes are not as susceptible to online cheating as the more standard lower-level classes. Many upper-level classes use specialized vocabulary and definitions that can vary even with the same course, depending on the instructor. Such vocabulary variations may also occur with introductory-level courses, but at a much smaller scale. Upper-level math (post calculus) and computer science questions also appear in third party online exam/homework assistance platforms, but these are more difficult for any platform tutors to crack and keep up with.

Specialized Vocabulary and Concepts

Upper-level courses frequently employ specific terminology and concepts that might differ based on the instructor or the curriculum.

Variations in vocabulary make it more challenging for third party online exam/homework assistance platforms to provide accurate solutions.

Computational Questions

While engines like Chat GPT and Wolfram Alpha can solve certain computational questions, exam setters should ensure that the questions are crafted uniquely to prevent direct solutions from these platforms.

Proof-based Questions

Questions that require proofs, common in many advanced math courses, and some computer science subjects, are generally tougher for both third party online exam/homework assistance platforms and Chat GPT to handle.

4. Coding Questions

Chat GPT has the ability to generate computer program codes. For questions requiring a computer code/program: Programs must run and produce an output and if the instructor tests the program, it should produce the same output. The instructor requires that students provide their codes in a format that can be copied and pasted. It is also required that the codes use only functions and protocols that were introduced in class. It is allowed to use new functions and constructs, but anytime such a usage is done, the new function/construct needs to be thoroughly explained with examples. Students are also informed that if their codes appear to be beyond the scope of the class’s programming skills, the instructor may call them in after the exam to fully explain how their code works. Prior to Chat GPT, the author would allow students to use any new code strategies that were not taught in class without restriction; indeed, students are encouraged to go beyond expectations and study more programming methods. What usually happens when Chat GPT writes a computer program (in whatever programming language you ask for), it will feel free to use ANY functions and methods that it sees fit. Sometimes the functions used require the installation of some additional packages in order to run. This makes it fairly easy to detect when a student has used Chat GPT. Most often the programs will not run at all (in the author’s machine learning course the software R was used) and/or, the program will use some constructs that are more advanced or simply different than were taught in class. Thus it should be required that the programs the student actually provide will indeed be executable (and they should provide the resulting output). If you were to ask a student who used Chat GPT to write their code to explain their code in case more advanced constructs are used, they will most often become flummoxed. In order to avoid the necessity of such encounters to verify whether the computer code was indeed the student’s creation, it is simpler to simply not accept solutions that violate either of the above directions (or perhaps say that such solutions can earn no more than, say, 25% of the point values) in such cases.

Course-Specific Assessments

In other upper-level courses such as Cryptography, Discrete Mathematics, and Graph Theory, it is vital to frame questions based on the specific teachings of the course. This not only tests students' understanding but also mitigates the impact of online test support services.

In conclusion, while advanced courses present their own set of challenges for online assessments, a keen focus on course-specific teachings and a proactive approach towards leveraging technology can help ensure exam integrity and fairness.

The rapid transformation of the educational sector, catalyzed by unforeseen circumstances such as the pandemic, underscores the importance of adaptability and forethought. With distance learning not merely being an option but a necessity, online assessment, once considered a subsidiary component of education, has gained paramount importance. Our exploration unveils a multifaceted reality: the sheer potential of online learning, the unprecedented challenges posed by third party online exam/homework assistance platforms and ChatGPT, and the undeterred perseverance of educators in ensuring academic integrity.

Our findings highlight an essential paradox. The very tools, like AI and internet platforms, that hold the promise to revolutionize and democratize education, also present profound challenges to academic integrity. As technology continues its relentless march forward, the academic community needs to be one step ahead, always innovating and strategizing to maintain the sanctity of education.

While third party online exam/homework assistance platforms exploit the vulnerabilities of the current online assessment system for profit, AI models like ChatGPT offer a glimpse into the future of information accessibility. They underscore the urgency with which educators and institutions must rethink and reshape assessment methodologies. The traditional paradigms of assessments, predominantly reliant on rote memory and standard problem-solving, may increasingly become obsolete. Instead, a greater shift towards analytical thinking, application, and synthesis may not only align better with real-world challenges but also be less susceptible to the shortcuts technology offers.

From introductory courses to advanced levels, the narrative remains consistent: there is no single foolproof method, no silver bullet. Instead, the solution lies in a combination of approaches—continuous adaptation, employing course-specific nuances, leveraging technology judiciously, and fostering an academic culture grounded in integrity.

According to the author’s experience, the most important aspect for an online class to improve is the participation and peer communication of the students. I have noticed students have a much higher reluctance to participate in an online class than in an in-person class. There are many plausible reasons for this, for example, knowing that the class will be recorded and posted online might make students more hesitant to speak up. I am working on better understanding and mitigating this problem.

As we forge ahead into an increasingly digital future, it is imperative to view these challenges not as insurmountable barriers but as catalysts, driving us towards a more resilient, adaptive, and effective educational paradigm. It serves as a stark reminder that in the evolving dance between technology and education, staying static is not an option. The future of education demands foresight, adaptability, and a relentless commitment to academic excellence and integrity.

Readers might be curious about how the author’s department and university are dealing with online courses with the learning experiences of the pandemic behind us. The author’s campus belongs to the California State University system, which comprises of 23 campuses throughout the state and is the largest university system in the United States. While particular policies vary by department and campus, the system’s higher administration, however, has made clear its preference to go back mostly to face-to-face instruction. At our college, in order to be offered online, any course must be approved by the department and the college dean, and then a university-wide committee. This academic year, our mathematics department has only two courses that have been approved for online instruction, one of which is the machine learning course that the instructor is currently teaching. A notable exception is our introductory statistics class, the main online course that was analyzed in this paper. This is the department’s largest course, with over 40 sections offered every semester. The author continues to teach the (only) large lecture for this class, but at present there are no online sections being offered. As discussed, online courses have many advantages for democratizing education, but at the same time, there are important assessment issues that need to be further examined, so this topic will be an important area of research in years to come.

Data availability

All data for this paper was included in the paper, either in raw form or when more suitable via a graphical summary.

Code availability

Not applicable.

Sleator RD. The evolution of eLearning-background, blends, and blackboard…. Sci Prog. 2010;93(3):319–34. https://doi.org/10.3184/003685010X12710124862922 .

Article   Google Scholar  

Bartolic SK, et al. A multi-institutional assessment of changes in higher education teaching and learning in the face of COVID-19. Educ Rev. 2022;74(3):517–33. https://doi.org/10.1080/00131911.2021.1955830 .

Oxford (University) Learning College webpage. https://www.oxfordcollege.ac/news/history-of-distance-learning/ . Accessed 15 July 2024.

Bernard RM, Abrami PC, Lou Y, Borokovski E, Wade A, Wozney L, Huang B. How does distance education compare with classroom instruction? A meta-analysis of the empirical literature. Rev Educ Res. 2004;74(3):379–439.

Horspool A, Lange C. Applying the scholarship of teaching and learning: student perceptions, behaviours and success online and face-to-face. Assess Eval High Educ. 2012;17(1):73–88.

Horizon 2020 TeSLA Project. https://tesla-project-eu.azurewebsites.net/ . Accessed 19 Feb 2024.

Fidalgo P, Thormann J, Kulyk O, et al. Students’ perceptions on distance education: a multinational study. Int J Educ Technol High Educ. 2020;17:18.

Gamage KAA, de Silva EK, Gunawardhana N. Online delivery and assessment during COVID-19: safeguarding academic integrity. Educ Sci. 2020;10:301. https://doi.org/10.3390/educsci10110301 .

Newton P, Essex K. How common is cheating in online exams and did it increase during the COVID-19 pandemic? A systematic review. J Acad Eth. 2023. https://doi.org/10.1007/s10805-023-09485-5 .

Noorbehbahani F, Mohammadi A, Aminazadeh M. A systematic review of research on cheating in online exams from 2010 to 2021. Educ Inf Technol. 2022;27:8413–60. https://doi.org/10.1007/s10639-022-10927-7 .

Dendir S, Maxwell RS. Cheating in online courses: evidence from online proctoring. Computers Hum Behav Rep. 2020;2:100033.

Golden J, Kohlbeck M. Addressing cheating when using test bank questions in online classes. J Acc Educ. 2020;52:100671.

Holden OL, Norris ME, Kuhlmeier VA. Academic integrity in online assessment: a research review. Front Educ. 2021. https://doi.org/10.3389/feduc.2021.639814 .

Jia J, He Y. The design, implementation and pilot application of an intelligent online proctoring system for online exams. Interact Technol Smart Educ. 2021. https://doi.org/10.1108/ITSE-12-2020-0246/full/html .

Nguyen JG, Keuseman KJ, Humston JJ. Minimize online cheating for online assessments during COVID-19 pandemic. J Chem Educ. 2020;97(9):3429–35. https://doi.org/10.1021/acs.jchemed.0c00790 .

Jeon J, Lee S. Large language models in education: a focus on the complementary relationship between human teachers and ChatGPT. Educ Inf Technol. 2023. https://doi.org/10.1007/s10639-023-11834-1 .

Download references

Apart from the author’s university salary; there was no additional funding for this paper.

Author information

Authors and affiliations.

Department of Mathematics, California State University-Dominguez Hills, 1000 East Victoria Street, Carson, CA, 90747, USA

Alexander Stanoyevitch

You can also search for this author in PubMed   Google Scholar

Contributions

A.S. is the sole author.

Corresponding author

Correspondence to Alexander Stanoyevitch .

Ethics declarations

Competing of interests.

The authors declare no competing interests.

Additional information

Publisher's note.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ .

Reprints and permissions

About this article

Stanoyevitch, A. Online assessment in the age of artificial intelligence. Discov Educ 3 , 126 (2024). https://doi.org/10.1007/s44217-024-00212-9

Download citation

Received : 31 October 2023

Accepted : 18 July 2024

Published : 19 August 2024

DOI : https://doi.org/10.1007/s44217-024-00212-9

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Online assessments
• Integrity of exams
• Pandemic-induced education
• Online cheating
• Artificial intelligence in education
• Online exam strategies
• Assessment challenges
• Student behavior

Advertisement

• Find a journal
• Publish with us
• Track your research