Skip to content

The promise of precision agriculture – can AI help to feed the world?

Franziska Bentz is a PhD Student in the Doctoral Programme in Biomedicine at the University of Helsinki. As someone who is equally passionate about science, writing and communication she strives to bridge the gap between science and the public.
This article is part of the trials, errors and breakthroughs theme.

edited by mimmu & heini, and illustrated by Vicky. This article has not been formally reviewed. Should you have any comments, please let us know!

Failing is a crucial part of human life. It is so crucial that without failing, we would not be able to appreciate the greatness of the successes that follow. One of the most astonishing breakthroughs in our lives is probably when we learn to walk. Learning to walk results from continuous trial and error: on average, toddlers fall no less than 17 times per hour! Each time, they get up on their feet again – and it is that getting up that makes it a breakthrough.

Now, imagine agriculture is like this toddler—humanity’s very own precious little baby with the power to nurture humans all over the world. Most experts would agree that the first step in agriculture’s incredible journey was the Neolithic Revolution, which started 10,000 years ago and lasted about 5,000 years. During this time, ancient communities worldwide transitioned from a nomadic hunter-gatherer society to agriculture and settlement. But what might look like a breakthrough from today’s perspective did not come about without failures.

Initially, agriculture meant harder work and provided less nutritious food than hunting and gathering. Domesticating only one or a few plants shifted the diet to more complex carbohydrates, opening the door to malnutrition and famine. Furthermore, the proximity to domestic animals increased the prevalence of plague and pestilence.

Since the Neolithic Revolution, agriculture has grown a lot, and we know more about the science behind it than ever before. Humanity has made great advancements in breeding plants with optimal character traits for our purposes and developed amazing technologies and machines to simplify the process of farming. But suppose this means that agriculture has finally grown up and is ready to fulfil the high hopes humanity is putting into it. Why is the threat of a food crisis still a repeating pattern in human history?

Studies show that we produce enough food to feed 10 billion people already. Does this mean that the food crisis is a problem of distribution – a political issue rather than a scientific one? Is agriculture bound to remain in its infancy forever? Or is the problem, instead, that humanity’s demands on agriculture are increasing much quicker than agriculture can ever grow? What exactly is this food crisis experts fear in the 21st century?

In 2022, the number of people suffering from hunger was estimated to be around 828 million worldwide. The food crisis results from a complex interplay of many factors. Already in 1999, the Nobel-winning economist Amartya Sen stated, “No famine has ever taken place in the history of the world in a functioning democracy”. Political will aside, among the most significant issues is food waste, which plagues 30-40% of the global food supply. This waste is perpetuated by an inefficient distribution system driven by both a lack of infrastructure and knowledge of how to keep food fresh. Moreover, natural events such as the COVID-19 pandemic and climate change aggravate this problem. The pandemic forced food factories to slow down or even stop their production, further disrupting the food chain. Simultaneously, climate change induced droughts, floods, heat waves, and wildfires, as well as shrinking access to water reserves and arable land, make it increasingly challenging for farmers to cultivate fruits.

By 2050, the population will reach nearly 10 billion people. Due to the increasing global demand accompanying this, agriculture is predicted to be required to produce 50 % more food, fibre, and biofuel compared to 2012. In other words, agriculture must nurture more people using fewer resources in more challenging conditions than ever before. Otherwise, even more people will be at risk of hunger and starvation. That sure sounds like a crisis science can help to overcome.

“Science prevented the last food crisis. Can it save us again?” asks an article in National Geographic about the next Green Revolution. To be able to answer this question, one first needs to understand the history of agriculture.

In addition to abrupt, courageous steps taken in the history of agriculture – such as the transition to settlement – some techniques have developed more gradually through improvements of existing systems. A great example is the three-field crop rotation system rooted in a one-field or two-field system used in medieval times. In the latter, the arable land was divided into two groups of fields, one of which was planted with rye, barley, or wheat, while the second lay fallow. Sometimes, livestock grazed on the unplanted areas, and their droppings enriched the nutrition content of the soil. The three-field system divided the land into three parts: one-third was planted with wheat, barley, or rye in autumn, and another third with oats, barley, and legume in spring. This means that only one-third remained unplanted, which almost doubled the crop yield. In addition, the legumes fixated nitrogen and improved the human diet at the same time by providing extra sources of protein. Albeit simple-sounding, this three-field system is considered a significant development in the history of agriculture, and it demonstrates the possibilities that come with increased knowledge and understanding.

The next major step in agriculture happened around 1913 when chemist Fritz Haber invented the method of synthesising ammonia from atmospheric nitrogen to use as artificial fertiliser. In addition to the use of high-yield crops and pesticides, these chemical fertilisers were one major driver of the Green Revolution that characterised the agriculture of the 20th century. Finally, they allowed food production to outpace population growth. For this reason, Haber was awarded the Nobel Prize for his invention.

However, nobody knew how greenhouse emissions affected the climate back then. As is nicely summarised here, ammonia manufacturing needs a lot of energy. Since it’s only partly taken up by the plants, ammonia either runs off into waterways or is broken down by microbes in the soil, which leads to the production of greenhouse gases. Solving one big problem has led to another, highlighting yet again how failure and success go hand in hand. For humanity in the 21st century, this now means that agriculture has to feed the world’s rising population while considering climate and environmental impacts – a tough challenge for humanity’s 10,000-year-old offspring.

To achieve the impossible goal of “doing more with less”, it is crucial that agriculture is once again willing to speed up its pace and employ the most recent developments of what some refer to as the revolution of the 21st century – the Information Revolution.

Nowadays, we can use various tools to collect data of different kinds, ranging from satellites, permanently installed cameras, and soil sensors to drones. Properly combined, these can provide extensive insight, for example, into the nutrition (such as nitrogen, phosphorus, and potassium) and water content of the soil, the size and ripeness of the fruits, or the emergence of disease both in a spatial and temporal context.

Evaluating and integrating all this information has led to the observation that each patch of farmland is a unique composition of dozens of measurable variables. Fascinating! But – if we are being honest – it’s not exactly helpful for farmers yet. How does this information-intensive map explain why different patches of farmyard vary in yield? Which concrete measures would a farmer have to undergo to maximise the crop yield while minimising the required resources?

Luckily, Big Data also comes with Big Opportunities. Various machine learning and deep learning algorithms have been developed to identify underlying patterns that are too abstract or complex for a human to understand. These technologies can advise farmers in all kinds of short- and long-term decisions like when and how to water or fertilise certain areas or which seeds to sow in the next season. In other words, precision agriculture is a data-driven soil and crop management system that allows farmers to cultivate their farms in response to inter- and intra-field variability. It sounds like agriculture has come a long way since its first steps.

But if agriculture has truly grown up, why are adoption rates so poor even three decades after introducing the concept of precision agriculture? A detailed summary points out the financial burden as one major contributor, especially given that benefits might not always be apparent immediately after implementation. The contributing factors also include connectivity issues, data management, and even literacy.

Norwegian farmer Lars Petter Blikom explains many more critical aspects in his blog series. In seven small yet very interesting articles, he describes what I would sum up as one major issue: the lack of cooperation between farmers and scientists. This starts with scientists not understanding the reality and perspective of farmers, along with the challenges they confront daily. Consequently, when scientists propose novel technologies, they may not align with the practical experiences or needs of the farmers. Unsurprisingly, they are then met with scepticism and trust issues from the farming community. A review summarising farmers’ views on implementing precision agriculture in France, the Netherlands, and Australia also observes insufficient systemic efforts, for example, by agronomists and the industry, to make precision agriculture work for farms. Specifically, they point out the problem of lack of software and hardware standardisation from different companies, insufficient support in implementing them, and the fear of being too dependent or losing flexibility when relying entirely on one system.

The 21st century is the century of the Information Revolution. The Internet of Things can help us build smarter cities, data-driven approaches have revolutionised modern health care, and large language models, like ChatGPT, are so advanced they could pass the bar exam. The pool of possibilities seems endless. Over the history of humanity, science has played a key role in the development of agriculture. So, it is hardly surprising that these technologies also find applications in precision agriculture with the great promise to “produce more from less”, i.e., to help nurture an ever-growing population with shrinking resources.

When first introduced in the 1990s, precision agriculture was expected to fundamentally change modern agriculture. However, as previously discussed, in 2024, the implementation of precision agriculture is still rather poor. Why have such groundbreaking inventions failed to find fertile soil in agriculture? Which direction does precision agriculture need to take in the future to meet the expectations of being the next revolutionary change, especially given that these expectations are growing as technology develops?

One of the interesting applications of precision agriculture is “variable zoning“, where data on the plot is fed into an algorithm, which divides the field into artificial zones based on shared attributes. Farmers can use this information to treat each of the zones in a unique way, adapting to the specific needs of that area. Unfortunately, current algorithms are still too simplistic and can, for example, only consider the amount of vegetation without considering that one area might have a nutrient shortage while the other is simply too wet. This means that to ensure the right approach is used, the farmer must manually check each area. And if the farmer must check first anyway, why would they invest in expensive sensors and algorithms? Additionally, farmers would need to buy the necessary machinery to work the field in response to the data. Imagine a farmer has identified different zones requiring different amounts of fertiliser. At this point, they would need a machine that can read the map and the corresponding amount of fertiliser, monitor its location, save the data, and transfer it back for documentation – for now, still “too complex, too much work and too many things can go wrong”.

Another problem with precision agriculture is the complexity and inaccessibility of both collecting and interpreting data. Technologies require an active cloud connection, but many farmers face frequent internet interruptions. Furthermore, current connectivity solutions are lacking when it comes to high-bandwidth data transmission from drones and cameras and pose an additional financial burden on farmers, contributing to the inaccessibility of precision agriculture. In some cases, a lack of standardisation prevents different tools from transmitting data between the machineries of different companies. At the same time, even when successfully acquired, data needs to be interpreted and understood before being turned into actionable information. Farmers all over the world report a limited support system from agronomists or companies in that domain, forcing them to either acquire the needed computational skills by themselves or form local networks with other farmers.

Another problem is that one of the highest cost factors is manual labour. Labour costs place a high financial burden on farmers and will continue to do so in the future. This highlights the need for technical advancements to automate labour-intensive aspects of production while overcoming the challenges of affordability and maintenance. Already now, robots are being used for tasks like tillage, planting, harvesting, and sowing seeds on some farms. While some robots are equipped with real-time precision spraying systems that drastically reduce the use of pesticides by applying the pesticide locally and only where needed, others are removing weeds (up to 200,000 per hour!) or insects with a laser, hence completely getting rid of chemicals. Amazing!

Looking back at the long life of agriculture, it becomes clear that humanity’s child has learned very well how to walk; it has gracefully gotten up after each and every fall and managed to overcome even the bumpiest roads. But in all the haste of learning to walk, humanity has failed to teach its child the second breakthrough in each individual’s life: to talk. In the future, the agricultural revolution may require agriculture to speak up for itself and tell science exactly what is needed for precision agriculture to deliver what it promised.

During my research, I found only one blog describing a farmer’s perspective on precision agriculture. One blog from one farmer in one country about a topic that is supposed to influence all farmers worldwide. While there are studies as well, most of them focus on only one country or area, often neglecting developing countries even though they make up 90% of farmers worldwide and will suffer most from the effects of climate change. This is not enough! To find universal, simple, and affordable solutions accessible to everyone, we need to hear the voices of every farmer – with different climate, financial, political, and personal backgrounds.

And when they speak, humanity must listen. In the long-shared history and interaction between science and agriculture, there is no doubt that the most successful agricultural revolutions have always resulted from a close collaboration between the two. But in the 21st century, the situation is more complex. We must consider the different perspectives, needs, and challenges of science and agriculture to ensure liveable conditions for each farmer and human today and in the future. We, that is, scientists, company owners, farmers, agronomists, scientific advisors, and every person using resources on this planet. Feeding the world has and will always be a problem that can only be solved when we combine every bit of human knowledge.