Unleashing the value of artificial intelligence in the agri-food sector: where are we? Open Access

Artificial intelligence (AI) is making rapid strides, playing a significant role in our civilization's growth and evolution. Having all the potential to revolutionize every aspect of the world, either in medicine, economy, leisure, communication, logistics and beyond (Apell and Eriksson, 2023; Bag et al., 2023; Malthouse and Copulsky, 2023; Tohidi et al., 2023), AI has gained more popularity along the diffusion of the fourth Industrial revolution and has become a perfect ingredient for every sector to flourish (Jan et al., 2022; Mariani et al., 2023).

AI is a vast field that invokes multiple facets, such as sensing, modelling, planning and action, as well as decision-support systems, natural perception, analytics and robotics (Camaréna, 2020). Its applications and its impact have been under investigation since 1950 (Nayal et al., 2022, 2023; McCarthy, 2004) when Alan Turing (1950) drove the attention to the subject by questioning “machines” ability to “think” in his research paper “Computing Machinery and Intelligence”. This impact of AI is also noticeable in sectors that are considered “traditional” as the agri-food one (Mônica da Silva Zanuzzi et al., 2020). In the last 30 years AI significantly contributed to shift the agri-food industry (Rejeb et al., 2022). AI has been used to accomplish operation efficiency (Liakos et al., 2018), increased food quality and quantity, promote sustainability (Bechar and Vigneault, 2016; Wolfert et al., 2017), as well as to support decisions (Deo and Şahin, 2015) and increase competitiveness and profitability (Slaughter et al., 2008; Xiong et al., 2017).

Currently, AI plays a key role in the success of various sectors, including more traditional ones, such as agri-food. Even if its potential is not fully understood, it is evident that AI has the capability to guide managers in adopting the best strategies and making well-informed decisions in every stage of the value chain (Soltani-Fesaghandis and Pooya, 2018; Dhamija and Bag, 2020).

The agri-food sector presents some peculiarities that make AI particularly appealing to companies along the entire supply chain. According to the United Nations' 2019 forecasts, factors such as climate change, world population (expected to reach 9.7 billion by 2050), food deterioration, and changes in consumers' food preferences have created the need for faster, more customized, and more efficient production. These reasons are the main drivers of the full integration of AI in the referred sector.

However, despite recent literature reviews on broader topics such as agri-food entrepreneurship (Petrolo et al., 2022), food value chain and open innovation (Misra and Mention, 2021), business models for vertical farming (Biancone et al., 2022), and the application of new technologies in consumer behavior studies (Berčík et al., 2021), knowledge about AI in the agri-food context remains insufficient. While some attempts have been made in computer science, where scholars have analysed extant research on AI implications in food from a digital transformation perspective (Monteiro and Barata, 2021), to the best of our knowledge, no comprehensive studies have been conducted until now in the management area. This knowledge gap is particularly intriguing because the potential of AI in the agri-food sector is central to several EU-funded initiatives (e.g. the Italian PNRR Agritech), and its benefits could have significant impact companies of all sizes operating the entire agri-food value chain. Therefore, we individuated the following research questions:

The remainder of this paper is structured as follows. In the next section, we begin by identifying and describing the main avenues of AI and their application in the agri-food sector. Following that, we provide a detailed overview of the methods employed in this study. Based on an initial analysis of 407 papers in the management field, we highlighted an intellectual core of 69 contributions. This intellectual core was then subject to a bibliographic coupling analysis to identify the current trends of AI in the agri-food sector. Subsequently, we conducted a further systematic review of all the documents. These analyses led to the proposal of an interpretive framework that aims to guide future research and inspire practitioners regarding the potential applications of AI along the agri-food value chain. Finally, the paper concludes with limitations and future research avenues.

AI has caught the attention of researchers over the years, with various studies analysing its contributions and applications to multiple sectors. AI affects our lives and societies whenever a computer is used (Mhlanga, 2021) and it has been a topic of debate since the invention of the Turing test (Baryannis et al., 2019; Dwivedi et al., 2019).

Several literature reviews and studies relevant to AI and its applications have been published (Toorajipour et al., 2021; Gao and Ding, 2022), which have shed light on various definitions of AI. McCarthy defined AI as “the science and engineering of creating smart machines and computer software to comprehend human intelligence” (McCarthy, 2004, p. 2). Spanaki et al. (2022) referred to it as a cognitive process, while according to Encyclopaedia Britannica, it is defined as the ability of computer-controlled robots and digital computers to perform tasks associated with human intelligence (Copeland, 2019).

Comprehending AI requires the proper understanding of its “soft” and “hard” sides, i.e. human intelligence (Camaréna, 2020) and Machine Learning (ML) respectively. ML is an Industry 4.0 (I4.0) technology (Nayal et al., 2022, 2023) and is often regarded as the most straightforward way to accomplish AI (Bowling et al., 2006).

One of the application sectors in which AI can be a valuable tool (Spanaki et al., 2021; Yahya, 2018) is agriculture (Mhlanga, 2021). Looking at the studies analysing AI applications in Agri-food, it is possible to recognise an evolutionary path moving from specific applications to an all-encompassing approach (Rejeb et al., 2022). This transformative impact has been further amplified by the emergence of I4.0, that has influenced the agricultural sector through from the integration of other enabling technologies. This has led to the emergence of agriculture 4.0 (Suma et al., 2017; Patil and Kale, 2016). Agriculture 4.0 plays a key role in generating and analysing data that becomes instrumental in making accurate decisions, ultimately, leading to enhanced productivity, efficiency, and sustainability (Araújo et al., 2021). While previous reviews have already highlighted the evolutionary trajectory of AI in the agri-food domain, our focus more on the underlying logic of AI in research works. Therefore, based on the literature reviewed, we can describe AI in a narrower and broader manner. “Narrow AI” regards specific intelligent technologies for a definite output, and even slight changes can generate different outcomes (Goertzel, 2014). “Broad AI” refers to advanced complex and highly adaptive intelligent systems characterized by efficiency and reasoning (Hochreiter, 2022). In the former, AI is an enabling technology linked to agriculture 4.0. In the latter, AI (particularly traditional ones such as agri-food, where it drives the change of business models, as well as business structural characteristics. The potential range of functionalities that AI could encompass within individual functions and across inter-functional levels may require changes or updates in the organizational architectures as well as in the competencies and skills required to be fully explored.

In conjunction with earlier literature reviews, we have noticed that the intersection of agri-food and innovation, including AI and smart technologies, represents a novel emerging topic since all the reviews date from 2021 or later. Table 1 provides an overview of these studies, which have all employed bibliometric and/or systematic analysis methods. Briefly, previous literature has tackled the topic from both tangible and intangible perspectives. In terms of tangible aspects, Monteiro and Barata (2021) identified the main research clusters of agri-food supply chain studies and conducted a detailed content analysis of 18 papers to ascertain the type of AI employed and its role in enhancing supply chain efficiency. They concluded that the impact of AI on food production is rapidly increasing and emphasised the importance of adapting these technologies across the entire supply chain in terms of stakeholders' engagement and exploiting multiple data sources to obtain sustainable agri-food (e.g. reducing food waste and increasing awareness) (Monteiro and Barata, 2021). Nonetheless, Biancone et al.’s (2022) study of 186 papers revealed what technological embracement holds, not only in terms of new agri-food market establishment in an urban context, but also in terms of entrepreneurship encouragement. Moreover, encouraging technological implementation could boost nutritional food quality and promote sustainability in the agri-food sector (Biancone et al., 2022).

Intangible aspects of agri-food and AI are represented in entrepreneurship and/or innovation-related studies. On one hand, Petrolo et al. (2022) through a bibliographic study and a systematic analysis, investigated the phenomenon of agri-food entrepreneurship concluding that agri-food entrepreneurs play a key role in developing ecosystems. Nonetheless, more efforts should be directed towards educating agri-food entrepreneurs on how to adapt different practices and develop their soft and hard skills to overcome sociocultural challenges (Petrolo et al., 2022). On the other hand, Misra and Mention (2021), through the deployment of a bibliometric analysis of 84 publications, emphasized that open innovation concurs with a growing space in food innovation research. The effects and challenges that open innovation holds along the food value chain are summarized through a framework (Misra and Mention, 2021).

This study relies on the combination of a bibliographic analysis and a systematic literature review of AI for the agri-food sector. To fulfil the objectives of this study, we passed through different steps. First, we conducted a search on the Scopus database, which relies on the identification of keywords. Scopus provides the largest number of scientific publications compared to its peers like Web of Science (WoS) (Gavel and Iselid, 2008). We used the following string TITLE-ABS-KEY (“artificial intelligence” OR “AI”) AND TITLE-ABS-KEY (food* OR agri*). The query was run in February 2023. We considered only publications up to 2022. Second, the results obtained were limited to the field of “Business, Management, and Accounting”, this limitation was set to only keep on journals in the field of management and business-related studies since, we seek to analyse the topic from a management perspective. Furthermore, to avoid translation overlap, we restricted the results only to English publications. The final output consisted of 407 publications, which form the initial core of this study. To focus on the most significant and representative articles related to this study topic, the authors applied a set of inclusion and exclusion criteria. They retained articles only if their title, keywords and abstract clearly provide insights about AI implications in the agri-food sector.

This screening process resulted in the identification of 129 pertinent publications related to agri-food and AI, which represent the intellectual core of this study, covering approximately 32% (31.9%) of the initial database. Over these 129 documents, we performed a bibliographic coupling analysis using VOSviewer software (Van Eck and Waltman, 2010). It allows grouping contributions based on the references they share; the more they overlap, the more papers are grouped in the same cluster (Casprini et al., 2020; Van Eck and Waltman, 2010).

Second, we conducted a systematic review of all the articles representing the bibliometric results. This resulted in 69 documents, covering the timespan 2009–2022. Following the systematic method proposed by Toorajipour et al. (2021), which draws from four phases (research question formulation, data collection, data evaluation and analysis, and interpretation) method designed by Denyer and Tranfield (2009). Our data analysis provides an overview of the paper topic, methodology, and research touch points for each of the 69 publications, as illustrated in Table A1 in the  Appendix.

Finally, based on both the bibliometric and systematic results, we provided an integrative and interpretive framework (reported in Section 5) linking the main streams that emerged, and providing guidance for future research and practice.

Reference-based analysis is the most common scrutiny method used (Gao and Ding, 2022). As shown in Figure 1, 69 publications were linked, forming 10 distinct clusters as shown in Figure 1. Based on an in-depth analysis, we grouped Cluster 2 with Cluster 4, and Cluster 5, Cluster 7 with Cluster 8 and Cluster 9 with Cluster 10, because of topic overlap. This resulted in the identification of six main clusters, as illustrated by Table 2.

Cluster A comprises 15 documents (timespan 2019–2022) which primarily focus on the adoption of AI and its benefits in the agri-food business. The papers explore how AI enhances customer engagement (Alfalih, 2022) and shapes their decision-making process (Tassiello et al., 2021). Additionally, they explore the role of AI in boosting entrepreneurship and innovation (Secinaro et al., 2022; Lachman and Lopez, 2019). Moreover, AI plays a key role in enhancing the adoption of sustainable practices in agriculture practices to meet sustainable development goals (SDGs) (Abban and Abebe, 2022; Morales and Elkader, 2020).

Cluster B, comprising 26 documents from cluster 2, 4 and 5 (timespan 2009–2022) and focuses on AI applications for efficiency and productivity enhancement. These AI-based technologies help predicting and monitoring crop diseases (Anand et al., 2022; Khan et al., 2022), livestock (Goel et al., 2022a), also they aid to forecast production such as in the case of Pakistani wheat (Ahmed and Hussain, 2022), to solve crop selection problems (Kumar et al., 2015) and to boost crop yield (Fenu and Malloci, 2021; Jain and Choudhary, 2022). However, sustainability and agriculture production efficiency are intertwined, as AI could enhance sustainable digital transformation (Lugonja et al., 2022) and farming practices (Alam et al., 2020).

Cluster C is composed of 10 documents (timespan 2010–2021) and centred on mapping supply and demand in terms of food logistics (Basljan et al., 2021), delivery management (Bocewicz et al., 2017), supply chain management (Sitek et al., 2017) and inventory management both at retail (Basljan et al., 2021) and warehouse (Lao et al., 2010b) levels. Nonetheless, AI-based technologies enhance sustainable food supply chains (Ting et al., 2014) and systems (Lao et al., 2010a).

Cluster D contains 5 items (timespan 2017–2022) represented by cluster 6 and focuses on decision support systems that assist consumers in making healthy food choices (Kim et al., 2021; Fiore et al., 2017), help firms to forecast sales (Dellino et al., 2018) and monitor their networks (Olan et al., 2022) to achieve sustainable supply chain management.

Cluster E, comprising 8 documents, is composed of cluster 7 and cluster 8 (timespan 2010–2022); it explores AI role in mitigating risks in the context of supply chain (Nayal et al., 2022), sewage sludge management (Valls et al., 2010) as well as managing these risks (Vuppalapati, 2022). This cluster also emphasises the significance of AI in addressing challenges faced by the agri-food industry such as climate change, resource scarcity, growing population (Camaréna, 2020); food safety and security, food demand and supply gap management, and food product quality (Dadi et al., 2021). It further supports the transition towards sustainable food systems (Brunori et al., 2022; Agyemang et al., 2022).

Finally, Cluster F has 5 contributions (2019–2022) resulting from cluster 9 and cluster 10, and it discusses AI marketing-related aspects from multiple perspectives. From a consumer viewpoint, it focuses on consumer perception of food quality as in the case of robotic chefs (Zhu and Chang, 2020) as well as understanding their preferences and attitudes towards organic food products (Taghikhah et al., 2021). From a firm's perspective, it examines price forecasting (Chu et al., 2019), and from retailers’ perspective, it concentrates on enhancing sales and promoting sustainability (Yang et al., 2022).

This analysis provides some general considerations. First, regarding methodologies, we remark that among the 69 analysed documents, most of them (37) are quantitative studies, while only around 20% of the total sample concerns qualitative ones (14) and only few works are literature reviews (7), conceptual papers (6) or have a mixed method analysis (5). These results explain that the implications of AI in the agri-food sector are an emerging topic currently under investigation. Second, multiple papers have highlighted the importance of driving attention to human-related aspects of AI.

In line with the objectives of this study, the systematic review identifies four primary research orientations as illustrated in Table A1. Firstly, most of the analysed documents investigate the application of AI technologies to improve efficiency, productivity, and ultimately, profitability. Previous research has explored the use of AI within the agri-food sector across various contexts, including smart farming, where it is employed to boost effective agricultural practices (Moreira et al., 2022), address challenges in suburban and rural areas (Doshi and Varghese, 2022), and bolster sustainable digital transformation (Lugonja et al., 2022). Additionally, AI has been used for predicting crop yields (Jain and Choudhary, 2022; Saranya et al., 2021), optimizing crop selection to improve agricultural planning (Kumar et al., 2015). Moreover, AI-based technologies and systems have proven effective in diagnosing and detecting plant and crop diseases, which can contribute to higher production levels (Moreira et al., 2022; Fenu and Malloci, 2021).

Secondly, the analysis of the available literature reveals a research trend centred on AI applications for external relations and stakeholders within the agri-food sector. This domain encompasses:

1. AI integration in agri-food supply chains, for its potential in enhancing sustainability (Ting et al., 2014), addressing perishable goods delivery challenges (Basljan et al., 2021), constructing resilient food networks (Bocewicz et al., 2017), and optimizing supply chain management (Sitek et al., 2017) and expenses (Nakandala et al., 2016).

2. AI solutions for marketing and communication, as they can improve agricultural e-commerce efficiency and competitiveness (Li and Xiao, 2021), identify attitude behaviour gap towards organic food product (Taghikhah et al., 2021), provide accurate price predictions (Chu et al., 2019), and optimize demand forecasting in retail (Priyadarshi et al., 2019). In this context, Lee et al. (2021) explore consumer intentions to use AI-powered food service stores, while Zhu and Chang (2020) examine the impact of robotic chefs on consumer expectations and perceptions of food quality. Hence, in the context of agricultural entrepreneurship, emerging technologies like AI hold potential to foster and promote agri-businesses (Secinaro et al., 2022).

3. AI tools to support decision making in agri-food, as they can assess future food systems (Agyemang et al., 2022), enhance fresh and perishable food supply chains management (Dellino et al., 2018) and support healthy nutrition decisions as it has the potential to deliver personalized nutrition management and dietary guidance (Kim and Chung, 2020).

Furthermore, the third pillar of the literature synthesises AI adoption in the agri-food business. Issa et al. (2022) offer a comprehensive explanation of AI readiness and identify strategic elements for agritech companies to ameliorate their technology adoption process. Ganeshkumar et al. (2022) explore the agritech industry digital transformation, highlighting the potential benefits and challenges of AI-based products in agriculture. Furthermore, Sood et al. (2022) identify the main drivers and barriers to AI adoption in agriculture.

Finally, a review of existing AI literature reveals a rising emphasis on various dimensions of sustainability, including environmental, social, and economic aspects. Sustainability is strongly intertwined with AI based technologies adoption to the agri-food industry. These technologies have been found to facilitate and to support sustainable digital transformation (Yang et al., 2020; Lugonja et al., 2022), promoting responsible and sustainable practices in the agri-food sector (Yang et al., 2022). In terms of social sustainability, AI can foster equitable growth in developing nations (Morales and Elkader, 2020) and enhance societal well-being (Ali and Xia, 2022). Within the realm of food security, which is a crucial component of both environmental and social sustainability, researchers contend that AI can ameliorate food security by addressing related challenges (Camaréna, 2020; Spanaki et al., 2022; Richards et al., 2009), optimizing agricultural practices and reducing waste. Furthermore, AI can optimize agricultural practices by integrating with renewable energy sources leading to increased efficiency (Doshi and Varghese, 2022).

In conclusion, AI adoption in the agri-food industry presents numerous opportunities and challenges. By understanding and addressing these challenges, the sector can harness the full potential of AI technologies to improve efficiency and productivity, to support agricultural sustainable practices and food security, to improve decision making processes, to promote entrepreneurship and to assure trust with various stakeholders.

This study serves as a comprehensive overview of the key international scientific contributions to AI and its applications within the agri-food sector, examined from a management perspective. Drawing from bibliographic coupling analysis and systematic literature review, this paper builds on recent contributions to the subject (e.g. Misra and Mention, 2021), and proposes an interpretive framework (Figure 2). The framework distinguishes three primary dimensions across which AI can be explored, as explained below.

We identified two sides of AI, entailing its technological aspects, which can be examined from two angles: the “hard” and the “soft”. The “hard” facet of AI focuses on technological development, adoption, and utilization. Research in this area aims at enhancing productivity and performance through AI technologies, thereby improving efficiency and effectiveness (Yang et al., 2020). A variety of AI implementations in the agri-food industry, ranging from crop disease detection (Anand et al., 2022; Darapaneni et al., 2022), to crop monitoring (Goel et al., 2022a), and crop productivity yield (El Hachimi et al., 2021; Kumar et al., 2015) and accuracy (Saranya et al., 2021) illustrate the potential of AI. Notably, the employment of robotic chefs to predict and ameliorate food quality as showcased by Zhu and Chang (2020) testify that AI holds a transformative power. Future studies could focus on deepening the knowledge of AI-driven technologies' role in business ecosystems (Yang et al., 2020), and understanding how machine learning could improve yield prediction (Murugesan et al., 2019).

On the other hand, the “soft” facet concerns stakeholders related sides of AI, emphasizing on the significance of the technology acceptance and embracement by consumers (Lee et al., 2021) and firms (Issa et al., 2022), an aspect that requires further understanding as highlighted by Zhu and Chang (2020). Nonetheless, voice-assistant interactions like “Alexa” are an intriguing domain to be deepened, shedding light on the effects of these interactions on the consumer experience and decision-making process (Tassiello et al., 2021).

An example of AI “soft” side is illustrated by the case of firms opting for recent AI-based technologies and exploiting their value to the fullest, assuring a high acceptance level by its individual and employees to adopt an adequate knowledge-transfer strategy should be incorporated. In this respect, the core analysis may be due to all human-related factors that facilitate or hinder AI endorsement. The literature on technology acceptance and diffusion (e.g. Rogers, 1962) could be helpful in guiding future research. Increasing the spread of innovative technologies and ideas in societies might change their food diets (Kim and Chung, 2020), as higher diffusion of innovation could shrink certain food product consumption and favour others and thus change the purchasing behaviour towards more sustainable consumption patterns (Fiore et al., 2017).

Secondly, we recognize two primary levels of analysis: (a) the firm and (b) inter-firm networks. Within the firm, the impact of AI can be categorised into two branches: (a1) the individual function, where AI is applied to a specific value chain activity (e.g. research and development (R&D)), and (a2) the inter-functional level, where AI impacts various organizational functions (necessitating top management to reevaluate the entire network of relationships and activities, both internally and with external partners).

For instance, in the context of the individual functions, Nayal et al. (2022, 2023) suggest that AI has a positive and significant role in managing agricultural supply chain risk. In contrast, at the inter-functional level, digital technologies such as AI and ML prove to be effective in executing various strategies and thus efforts that span multiple value chain activities (Ali and Xia, 2022).

At the inter-firm level, we have identified two types of AI-influenced networks. Firstly, the backward network which encompasses connections with suppliers. In essence, AI-based technologies can boost performance and co-create value among various parties, including suppliers and distributors of AI (technology providers) that either directly or indirectly contribute to the development and delivery of products and services (Yang et al., 2020). Secondly, the forward network pertains to a firm's relationships with customers and consumers, such as in the case of consumer voice-assistant interactions (e.g. “Alexa”), which prove advantageous (Tassiello et al., 2021).

We aimed to identify the potential impact of AI on all agri-food value chain activities. For simplicity, we relied on extant frameworks of the agri-food value chain, distinguishing among R&D, production, processing, storage and distribution, sales and pricing, consumption, waste disposal, and recycling (van der Vorst et al., 2001). For example, Kumar et al. (2015) introduced a crop selection method and proved that it could ameliorate the net yield rate of crops over seasons. Another example may relate to the adoption of new technologies to reinforce the user experience. Zhu and Chang's (2020) study of the anthropomorphism effect of robotic chefs on quality prediction highlights that real consumption experiences involving diverse cultures should be studied, and robots with higher levels of anthropomorphism should be examined to deepen our understanding of its impact on food quality.

Moreover, AI can make significant contributions to sustainability in the agri-food sector. It can be effective in addressing both environmental and social challenges at a niche level of agri-food businesses as well as to the broader ecosystem, shaping the future of the agri-food sector. The environmental aspect of AI revolves around resources management (Alam et al., 2020; Camaréna, 2020) and optimizing their utilization processes (Rahutomo et al., 2019), waste management (Yang et al., 2022) and employing renewable energy to enhance sustainable practices in the sector (Doshi and Varghese, 2022). These elements are interconnected with AI potential to enhance productivity and efficiency in the sector.

Simultaneously, AI can contribute to social welfare (Ali and Xia, 2022), ensure food security (Camaréna, 2020; Richards et al., 2009) and foster sustainable development goals like zero hunger and sanitation (Morales and Elkader, 2020).

This study reviews the literature on AI in the agri-food sector from a management perspective. We have identified six clusters to provide insight into the current research focus: (1) AI adoption and benefits, (2) AI for productivity and efficiency, (3) AI for logistics and supply chain management, (4) AI for supporting decision making process for firms and consumers, (5) AI for risk mitigation and (6) AI marketing aspects. This study adds to previous studies that have recently investigated multiple phenomena in the agri-food sector (Misra and Mention, 2021; Petrolo et al., 2022). Furthermore, we performed a systematic literature review in which we analysed 69 publications. In conclusion, we present an integrative framework aimed at providing better comprehension of AI integration in the agri-food industry and guiding future research to better analyse and develop topic-related contributions.

We argue that AI involvement in the agri-food sector drives value through three main dimensions. The first dimension entails two sides: the “hard” and the “soft”. The “hard” side focuses on technology development, adaptation, and usage, while the “soft” side emphasizes stakeholders' embracement and acceptance of these technologies. The second dimension covers the firm and inter-firm sides; the former reflects AI's impact on specific activities of the value chain (individual function), the latter refers to its influence on multiple value chain activities (inter-functions). The inter-firm level pertains to AI-influenced networks, which can be backward (with suppliers) or forward (with consumers). The third dimension is conclusive and concerns the role, applications, and effects of AI on the value chain process. We conclude that sustainability is a transversal out of AI integration in the agri-food sector, which links all the mentioned dimensions.

Based on the in-depth systematic literature, we believe that future research should focus on defining the theoretical underpinnings of AI in the agri-food sector. For example, we recommend that scholars consider theories related to ethics (as one of the key challenging and interdisciplinary areas of inquiry), consumer experience (as one of the main marketing aspects that could be affected by AI), and quality and sustainability (as core to production-related studies).

There are three primary challenges that must be addressed. The first is communicating the concept of AI and its full benefits to agri-food sector participants, including farmers, entrepreneurs, and stakeholders. A gap exists between the analysed literature and practitioners' needs, while academics and experts exploring cutting-edge applications of AI, farmers first require basic knowledge and understanding of AI's real advantages and benefits. Future research should ensure the creation of a common reference, based either on explicit knowledge (knowing-that, embedded through a series of trainings), to facilitate AI integration, or on tacit knowledge (knowing-how, gained with experience) to construct an intuitive wisdom-based model (Nonaka and Takeuchi, 1995). This model has the potential to strengthen AI implementation in the agri-food sector, overcoming challenges related to both the 'hard’ and ‘soft’ side of AI identified by this paper. Future research should also aim to develop a frame model to facilitate technology integration and understanding for stakeholders who may struggle with complex academic studies or scientific models. For example, future research could focus on AI utility to minimize human intervention from crop production processes (Charania and Li, 2020) to the evaluation of food quality prepared by smart robotic chefs (Zhu and Chang, 2020), also considering cultural differences (Tassiello et al., 2021).

The second challenge is about adapting and combining AI strategies in a more complex and dynamic environment, where traditional solutions and innovative strategies must find a new equilibrium by incorporating new knowledge. Within agri-food firms, AI can impact specific activities or multiple functions, necessitating updates that invokes various parties; thus, the establishment of a routine that structures new AI-based knowledge and guarantees its intertwinement with the existing structures and processes is deemed crucial (Filippini et al., 2012). Consequently, new research should not only identify novel solutions needed, but also outline new managerial profiles that can enhance knowledge absorption capacity by setting a common ‘frame of reference’ at the inter-firms’ level, between technology providers (suppliers) and technology receivers (buyers) (Filippini et al., 2012).

The third one refers to ethical concerns surrounding AI, a domain that requires additional research (Camaréna, 2020; Maulana et al., 2022), also with a multidisciplinary approach (e.g. law). For example, future studies could focus on developing business models that ensure proper privacy settings, data control, and confidentiality in the agri-food industry, particularly within the farming sector (Spanaki et al., 2022). Moreover, the supply chain and AI should be given more space in research, considering supplier behaviour towards innovative technologies (Olan et al., 2022).

This study has some limitations. First, we focused on a specific technology in a restricted sector, hence selecting a small core of documents. This could be subject to criticism. Second, we did not provide a large variety of future research avenues. Of course, there could be several other potential future research streams, but we want to enlighten the two main (and holistic) pathways for which we hope for a prompt and significant effort from scholars and experts in this field. Thus, after academics will cover the proposed challenges, further research can be made with a higher focus on more punctual problems and issues.

This research has partially received funding from the Horizon 2020 Programme of the European Union within the OpenInnoTrain project under grant agreement no. 823971. The content of this publication does not reflect the official opinion of the European Union. Responsibility for the information and views expressed in the publication lies entirely with the authors. A previous version of this paper was also presented at the 15th Annual Conference of the EuroMed Academy of Business: The authors thank the referees and the audience that provided insights for improvement. Part of this study was also conducted within the Agritech National Research Center and received funding from the European Union Next-GenerationEU (PIANO NAZIONALE DI RIPRESA E RESILIENZA (PNRR) – MISSIONE 4 COMPONENTE 2, INVESTIMENTO 1.4 – D.D. 1032 17/06/2022, CN00000022). This manuscript reflects only the authors’ views and opinions, neither the European Union nor the European Commission can be considered responsible for them.

Since acceptance of this article, the following author(s) have updated their affiliation(s): Meriam Trabelsi is at the Department of Economics and Management, University of Pisa, Pisa, Italy.