3.1 Sustainable Development and Artificial Intelligence (AI)

Sustainable development is a concept that has been gaining relevance in international politics, given the growing visibility of the environmental damage of the prevailing production model, represented, among other phenomena, in global warming, climate change, and the loss of biodiversity resulting from human activities, with emphasis on the burning of fossil fuels. In this sense, sustainable development is understood as the need to constitute a model of economic-social development that reconciles economic growth with the balance of ecosystems, intending to protect the planet's natural resources.

The term sustainable development is perceived as the possibility of enhancing the quality of life through the improvement of the human-environment relationship, in which the survival of ecosystems is guaranteed¹⁰. It involves a series of systems or dimensions: political (participation of actors), social (equity in distribution), cultural (consideration of local culture), technological (innovative solutions), productive (efficient use of resources), ecological (ensuring the balance of ecosystems and diversity), and international (sustainable trade models), among others. Therefore, it is considered a complex notion, since it involves multiple aspects that go beyond the responsible use of natural resources, moving towards a true balance between different systems ^{₂₎₍₁₀₎} .

In this way, sustainable development is constituted as a holistic category integrated by various dimensions, where ecological sustainability, social equity, and the global-local relationship acquire relevance, in addition to democratic participation and joint action on the different problems generated by climate change to promote the preservation of the planet and the quality of human life².

Given its importance, in 2015 all members of the United Nations adopted 17 goals with 169 targets, which include eradicating hunger, achieving food security, guaranteeing healthy lives, ensuring access to water and energy, promoting sustained economic growth, adopting urgent measures against climate change, among others⁶. In this way, sustainable development appears as a strategy to meet present needs without compromising the capacity of future generations, implying the need to balance economic growth, environmental protection, and social welfare.

This perspective is gaining strength in the need to establish a sustainable and productive model for the agricultural sector at a global level, where a new relationship is established with the planet and territories, ensuring a sustainable management and use of soils, water, and plant genetic resources, which are essential for food production⁵.

Moreover, by promoting sustainable development, the conservation of natural resources is supported, pollution is reduced, and social and economic equality is promoted, guaranteeing the long-term health of the planet. For this, innovation and efficiency in the use of resources in all economic areas are necessary. Thus, it is a great challenge for the agricultural sector, on the one hand, to increase food production to mitigate world hunger, but also to reduce the use of resources such as water, energy, fertilizers, and others.

These pressing needs are crossed by the accelerated technological development of recent decades, focused on the automation of various human practices, including the activities derived from agriculture. These technologies have driven the evolution of all production processes emerged from the so-called Industry 4.0, highlighting the transition from Agriculture 1.0 to 4.0, with predictions already pointing toward Agriculture 5.0 based on the permanent incorporation of innovations, whose advances are more vertiginous than in previous periods⁷.

One of these is precision agriculture (PA), which seeks to optimize crop yields through technologies that enhance accuracy in agricultural practices and decision-making. It is highlighted that PA incorporates sustainability considerations such as soil conservation, rational use of water, among others¹¹. In fact, by using technological applications, PA reduces the use of pesticides, fertilizers, and water, while promoting the optimization of soil fertility and crop yield. The aim is to make the use of agricultural resources more efficient, thus promoting a model of sustainable development⁹.

Within PA, reference is made to artificial intelligence, as its use has demonstrated the possibility of optimizing crop management, improving efficiency in the use of resources, preventing and controlling diseases, and providing accurate real-time data for decision-making. This enables more efficient crop management through the application of specific treatments and precise use of resources such as water and fertilizers¹².

Among the most prominent technological applications in PA are artificial intelligence, robotics, unmanned aircraft or drones, high-performance computing, and the Internet of Things. These innovations can increase efficiency and, at the same time, promote environmental sustainability by reducing the use of resources. Notably, they enhance the entire process from cultivation to commercialization¹².

Artificial intelligence, in particular, is understood as “a combination of technologies that brings together data, algorithms, and computing capacity. Advances in computing and the increasing availability of data are, therefore, a fundamental driver in the current pronounced growth of artificial intelligence”¹³. Every day, new advances emerge, and the forecasts in this field are impressive.

From this perspective, AI refers to machines with cognitive functions similar to humans, which manage to learn, understand, reason, make decisions, or interact. Therefore, AI systems integrate recognition and understanding of meaning, signal detection, intelligent control, and simulators through the combination of hardware components (sensors, chips, robots, etc.) with algorithms and software. As described, AI is a diverse and continuously growing field ^{₇₎₍₁₃₎} .

AI is thus a rapidly growing discipline whose promises even tend to defy science fiction. In agriculture, the incorporation of AI systems provides predictive information related to agricultural practices. For instance, AI can determine optimal sowing and harvesting dates; by enhancing crop performance, it reduces the use of resources such as water, fertilizers, and pesticides, thereby improving both efficiency and sustainability ^{₉₎₍₁₄₎} .

These solutions have been applied in various experiences, enabling agriculture to produce more with fewer resources. This explains their popularity and rapid extension in the world, especially in developed or developing countries, which report concrete experiences of application and diversification in agricultural environments. AI is considered an emerging field in agriculture that encompasses multiple research and development disciplines such as computing, machine learning, crop improvement, genetics, biology, statistics, physics, and chemistry, among others¹⁵.

For this reason, AI opens up an unprecedented range of options, focused on reducing inputs and automating tasks⁴. In the agricultural sector, specific fields of action stand out, such as crop yield prediction, price forecasting, intelligent spraying, agricultural robotics, crop and soil monitoring, with interaction with the environment being essential in the application of this type of tool ^_16)(17) .

It should be noted that among the AI technologies applied to agriculture, machine learning stands out. It applies artificial intelligence through a set of algorithms capable of executing complex tasks without requiring explicit programming. The objective is to feed the machine with data from past experiences and statistical information so that it can perform the assigned tasks and solve problems ^{₉₎₍₁₈₎} . For example, artificial neural networks in agriculture offer the possibility of predicting and forecasting based on parallel reasoning without continuous programming. They can be trained and can compensate for failures in expert systems by improving prediction methods¹⁸.

In this way, artificial intelligence has revolutionized agribusiness, since, for the first time in the history of humanity, the availability and use of information are enabling the optimization of production processes to reduce input investment¹⁹. Additionally, AI is also proposed as having the potential to develop sustainable agricultural processes.

3.2 Characteristics of Artificial Intelligence Applied to Agriculture

An interesting aspect of the transformations of agricultural practice is its strategy, which combines local agricultural knowledge with technological applications such as the massive acquisition of data and images, artificial intelligence, robotics, and task automation, thus facilitating intelligent decision-making to generate precise actions. Technologies in agricultural processes involve “the use of the Internet of Things, data analytics, artificial intelligence, machine learning, cloud computing, and remote sensors, which have been identified as the ones with the greatest use and impact on agriculture for water management, crop monitoring, climate analysis, agrochemical use, soil assessment, and greenhouse cultivation, among others”²⁰.

Likewise, within the AI techniques applied to agriculture, neural networks, machine and deep learning are noteworthy, allowing the measurements of crop production rates and providing information on supply chains. In addition, with the use of satellites, AI guides the detection of growth patterns in planted areas when applied to PA for crops, forests, or urban green areas, thereby providing strategic insights for effective decision-making⁴.

One of the innovations from machine learning and analytics has enabled the expansion of the Internet of Things (IoT), so that the data produced by these technologies feed back into them, allowing access to large datasets that can be managed more quickly¹¹. For example, machine learning is applied to “disease identification, yield prediction and biomass estimation; with prospects to develop simulations of crop growth and yield, machine training models from satellite data available free of charge”²¹.

Machine learning also enables machines to learn to extract patterns and relationships from data, improving their predictive abilities and decision-making. In agriculture, this includes predicting crop adaptability to soil and environmental conditions, while in livestock, it is used to automate animal tracking, minimizing losses and costs²².

Likewise, ongoing experiments are currently testing the use of robots and artificial intelligence for harvesting, drones or RPAS (Remotely Piloted Aircraft Systems), as well as other tools and instruments to digitalize agriculture. The implementation of sensors and predictive algorithms is also being tested to obtain data for crops and livestock management, with applications also extending to product commercialization ^_15)(18) .

Another area of AI are conversational tools, which combine natural language processing (NLP) with IoT devices based on advances in neural networks, which have become attractive, affordable and feasible devices for personal use in agricultural production⁸. Within the application of conversational AI in agriculture, experiences with the language model based on artificial intelligence ChatGPT stand out; “it is a sensitive artificial intelligence model, capable of learning from the preferences and behavior patterns of users by personalizing their response, even giving advice to personal questions”, with strong potential to positively transform agriculture based on proper information management²³.

In this order of ideas, AI in the agricultural sector automates various basic activities across the production chain: planting, harvesting, and crop, climate, and soil monitoring through learning algorithms to process data. Big data techniques and connectivity within an agricultural network (drones, sensors, software) also intervene, registering, storing, processing, and analyzing large amounts of data. Its central objective is thus to ensure optimal decision-making, minimizing uncertainty and, consequently, risks⁴.

From the heterogeneity that characterizes AI tools in agriculture, image processing, machine learning, and robotic systems are the most widely used and with the greatest growth forecasts²³. These advances have shaped the reality of agriculture 4.0, constituted from an articulated use of data from sensors, agricultural machinery, satellite images, drones, smart phones, and radio signals to convert them into useful information for crop management.

In short, AI applications are making it possible to increase productivity, reduce costs, and reduce the environmental impact of agricultural practices⁴. Some AI tools stand out for their agricultural improvements, such as remote sensors for detecting soil moisture, temperature, and pH, as well as automated irrigation guided by GPS. But, in practice, they often face challenges related to farmers' knowledge and implementation capacity¹⁴.

The implementation of robotic systems and artificial intelligence is also pointed out, since the robotic arm has proven effective at harvesting fruits and vegetables without damaging the plant, as well as selecting seeds, reducing failures, costs, and time, thus meeting productivity standards. Likewise, the use of autonomous robots for weeding is also highlighted, based on technologies such as GPS and drones, to detect signals and transmit instructions for the execution of weed treatment, with effective reports to increase accuracy²⁴.

As demonstrated, advances in AI, understood as the ability of a machine to perform activities associated with intelligent beings, have led to the constitution of programs capable of operating like humans, with their own rational way of processing information²⁵. In agriculture, AI has a preponderant role within digital agriculture or 4.0. This involves the implementation of AI algorithms and models in various areas, such as crop monitoring, irrigation system management, pest and disease detection, production optimization, and data-based decision-making. While these tools offer intervention options, it is still the farmer who makes the final decision; in short, a strategic integration between farmers and technologies is proposed.

The logic of AI technologies in agriculture seeks to improve the resources needed for production, not only water, soil, fertilizers, and pesticides, but also labor management. AI supports crop selection, the development of more resilient seeds, the optimization of irrigation and fertilization systems, and the identification of the best harvest times through autonomous robots and drones²⁶.

The application of such innovations in activities derived from agriculture has undoubtedly brought several benefits, including improved decision-making, effective use of information to optimize agricultural production, detection and prevention of plant diseases, abstention from accurate forecasts for climate management, and even supply chain management²³. Together, these advances increase yields, reduce costs, and minimize environmental impacts.

It should be considered that AI applications to agriculture are not limited to production, they also involve consumers, from the so-called “smart cities”, which facilitate product access, or the also mentioned “smart farming”, which aims to enhance resource efficiency through feeding or irrigation systems that use software to provide high-quality and sustainable products¹².

For these reasons, AI has become increasingly significant in agriculture, offering major opportunities to improve productivity, efficiency, and sustainability, given that “(AI) refers to computer systems with the capacity to perceive the environment, learn from experience, process large volumes of data and make decisions based on the information received”⁹.

3.3 Sustainable Development in the Use of Artificial Intelligence in Agriculture

As with any transformation generated in production processes, artificial intelligence presents a series of challenges. From the reviewed literature, a series of concerns emerge, including the need to promote sustainable increases in agricultural production, from its strategic integration with biotechnology, robotics, big data, simulation, and geostatistics¹⁹.

Despite the above, although innovations generated by AI tools in the agricultural sector are highlighted, they are valued as incipient when compared to other areas of the economy¹⁴. It is expected that the applications of these technologies in agriculture will continue to diversify, gradually incorporating them into all the activities that constitute agricultural practice worldwide.

It is important to mention that technological development in this field is not static; rather, it responds to an almost ephemeral dynamism, so it is essential to recognize the growing importance of data and how it is used. Due to the great benefits and implications AI has over the transformation of productive practices, further research is needed to assess its uses, advantages, disadvantages, and to identify opportunities for improvement. This, in turn, requires an adequate regulatory and ethical framework, which is not yet fully defined, although advances have been made, especially in Europe. Because of the aforementioned dynamics, ongoing evaluation is necessary²³.

Several key actors are identified in the process, on the one hand, the industries that produce these technologies and farmers who apply them, and on the other hand, governments, universities, and research centers, since not only knowledge of technologies and agronomy is required, but also infrastructure, legislation, knowledge, and even regulations on privacy and ownership of the data generated²⁷.

Most applications and experiences with AI, such as machine learning, have the potential to improve productivity and efficiency in agriculture and livestock, since they allow contextualized decision-making and efficient crop management through modeling strategies, classification, decision-making, and price predictions. In addition, studies emphasize socio-environmental benefits with an encouraging impact⁹.

AI in agriculture is therefore seen as a valuable tool for advancing sustainable development ^{₉₎₍₂₃₎ 25}. “Artificial intelligence is involved in the agricultural sector not only to benefit farmers but also to reduce pollution, care for and preserve natural resources and ecosystems by facilitating accurate information processing”²⁷.

Techniques such as image processing, machine learning, and environmental interaction appear as feasible approaches aimed at addressing the major global problems derived from population growth, climate change, and deforestation, helping to build an AI-driven agricultural system based on environmental preservation¹⁶.

Emphasis is placed on the benefits of technological applications, including AI-driven robotics, used to reduce farmers’ workloads, increase crop yields, expand and diversify crop types, and thereby strengthen food security. It may also reduce rural-urban migration, due to the creation of new companies and work practices. The intensification of sustainable production through the application of technologies such as no-till, mechanized weeding at the individual level, or very low-volume spraying, leads to greater production with fewer resources, thereby minimizing the use of inputs, soil alteration, and increasing food production without destabilizing ecosystems²².

This review highlights that the implementation of AI technologies creates the need to build the necessary capacity for the adoption of agricultural practices, raising concern among producers, since it demands higher levels of specialization ^_25)(26) . Resistance to the application of virtual tools is considered a natural phenomenon in the execution of any technological innovation, especially given that many agricultural processes have been carried out in the same way for almost a century.

From an environmental perspective, concerns are focused on reducing the carbon footprint and the efficient use of resources. Automated seed selection, for example, reduces losses caused by poor selection and optimizes the use of waste through proper classification²². According to Sachithra and Subhashini, the use of AI in agriculture has the potential to boost the achievement of sustainable development goals by improving farmers’ life quality, minimizing the cost of food production, regulating food prices, and ensuring food availability for consumers. However, they note that the goal of alleviating poverty within the farming community remains largely unaddressed²⁴.

In fact, in the face of critical positions regarding the application of technologies in agro-productive processes, it is estimated that AI can contribute significantly and effectively to achieving the Sustainable Development Goals¹⁷, but to do so, it is necessary to establish public-private partnerships, specifically between industry, academia and government, both to promote the development of technologies and to establish systems of vigilance in the generation and use of these innovations.

Another aspect pointed out is the management of large volumes of data, which implies the need for clear legal frameworks to ensure their security¹². Effective control of resources such as water, chemicals, and energy, while minimizing environmental impacts, also depends on the development of policies and strategies for monitoring and evaluation.

A relevant aspect raised by the European Union is the importance of human supervision in the application of AI tools. From that perspective, human autonomy over machines is emphasized, hence the importance of a control system and regulations in this matter. “The goal of trustworthy, ethical, and anthropocentric AI can only be achieved by ensuring adequate human participation in high-risk AI applications”¹³.