AI as a driver of agricultural transition

As a researcher, Athanasiadis has worked with computers for over two decades. He has seen his professional field change considerably over the years. “When I first started out, it was still all about humans teaching computers. You gave the machine a set of rules or examples and it then drew conclusions and made suggestions based on this information. Today – thanks in part to AI – that situation is completely reversed. Computers are now helping us to devise and find solutions, often at lightning speed and extremely accurately too. I find that fascinating. But that doesn't mean that human beings can now take a back seat. We are ultimately responsible for how and for what purposes we use computers and AI. That is something we have to consider carefully.”

With AI in a gold rush

Athanasiadis calls the rapid development of AI a ‘gold rush’, in reference to the massive influx of people of people to areas where gold had been found. “You see tech companies, big and small, launching more and more new AI tools at a faster and faster pace, in an attempt to show what we can do with AI. That is quite disruptive in some areas; think for example of ChatGPT. Both the industry, government agencies and public bodies are searching for the right balance to manage and regulate all this responsibly. Ultimately, this will allow us to deploy AI in a much less disruptive, safer, and more valuable way.”

AI in WUR research areas

AI currently plays a role in several WUR research areas, says Athanasiadis. “Think of plant phenotyping, in which researchers select specific traits from plants to develop new, better varieties. Thanks to AI, traits can be measured much faster and more effectively. Other examples include AlphaFold, a programme that can predict protein structures, and several AI applications in remote sensing and agricultural robotics.” Yet there is also still a lot to be gained when it comes to leveraging AI in WUR's research areas. Athanasiadis: “Among other things, we still lack good data benchmarks and standardised models for collecting, sharing, and processing data. That is what we hope to contribute to with our chair group.”

Developing concepts and methodologies

The chair group hopes, in collaboration with fellow researchers inside and outside WUR, to develop concepts and methods that experts in the various research areas can use in a smart and responsible way to develop AI tools, says Athanasiadis. “This might include tools that provide a better understanding of problems caused by climate change, but also tools that contribute to solutions to these issues, such as climate resistant crops, or biodiversity restoration. At WUR, we have a strong network of academics who can act as a link between research and the practical applications of AI. Having this combination under one roof is what makes WUR unique.”

Multidisciplinary approach

To work on global challenges and solutions, Athanasiadis says it is important for research to involve as much collaboration between different disciplines as possible. “So far, AI researchers have tended to work independently from domain scientists. One group was responsible for data collection, while another focused on data processing and analysis, and yet another worked on building an AI tool. Whereas you can devise much better solutions if you do it all within a single multidisciplinary team. I am not only talking about disciplines within WUR, but also about collaboration with parties in the professional field, such as farmers, companies, NGOs, and knowledge centres. Fortunately, WUR has a lot of experience with that.”

Involved in AI projects

For now, the AI Chair Group is mainly linking up with ongoing projects in which WUR is involved. Athanasiadis: “One such project is the European agrifoodTEF. In this project, we are working, among other things, on a digital infrastructure and facilities for small-scale companies to test AI-driven agricultural robots. Another project is LTER-LIFE. In this project, we build digital twins (digital copies of reality, Eds.) of ecosystems, using computational tools to understand the changing interactions within ecosystems due to climate change. Our goal is to eventually launch our own projects within our chair group, specifically aimed at developing AI methods.”

Preparing students for professional practice

Athanasiadis stresses that the chair group was not only created to conduct research in the field of AI. “We also have an educational task. We want to equip WUR students and graduates with good skills in the field of AI so that they are ready to enter the professional field. This means teaching them to use AI in ways that help them solve the right problems. To do that, you need to understand AI methods and know what is going on behind the scenes of AI tools. Clearly, we don't need all our students to become AI experts. You don't need to be a mechanic to drive a car either. But in today's digital world, basic knowledge of AI is a prerequisite.”

AI for social purposes

One of Athanasiadis’ ambitions as an academic scholar is to deploy AI not only for commercial interests, but also to serve society. “Unfortunately, funding for research in the field of AI is now mainly driven by fast commercial returns, i.e. that an AI application must eventually make a lot of money. As a result, current research often devote less attention to AI in relation to themes such as sustainability, climate change, biodiversity, and food security. I see it as the task of NGOs and universities to juxtapose this with a research agenda focused on applications that primarily serve people, nature, and the environment, and that are not purely financially driven. This is also an important reason why WUR is investing in AI in the form of a chair group.”

Not a replacement, but support

In conclusion: how does Athanasiadis envision the future of AI? In 10 years or so, will a farmer or grower be able to sit back on the sofa and watch Netflix while smart machines do the work in the field or greenhouse? Athanasiadis: “Certainly not; that image strikes me as quite dystopian. AI is not meant to replace humans, but to support them with data and information to make decisions, for example by advising on risk management or more efficient cultivation methods. If we use AI smartly, it can be of great value in accelerating the transition towards sustainable agriculture, and helping us become more resilient to climate change.”