“AI is a Must-Have for Anyone Who Wants to Do Top-Level Research” AI Think Tank “AI is a Must-Have for Anyone Who Wants to Do Top-Level Research”

Professor Bauckhage, the University set up the AI Think Tank just over a year ago as part of its Digital Strategy. What’s it aiming to achieve? 

The AI Think Tank brought together more than 20 colleagues from all manner of different areas, from computer science, law, medicine, philosophy and linguistics, but also from the administration and University IT. We had been asked to come up with recommendations for the Rectorate as to how the University of Bonn should be tackling the challenge of AI in the 21st century. We’re living at a time when machines are performing at an astonishing level and students are making active use of these systems to write their term papers, for instance. A whole range of different views from the various disciplines informed our thinking in the Think Tank. For the lawyers, for instance, the focus was on privacy, data protection and liability, while philosophers wanted to look more closely at the nature of intelligence and ethical considerations. We all complemented one another really well in order to cover as many aspects as possible. As a computer scientist, I’m definitely in the optimistic camp: AI will be able to make life easier for us in many ways if we use it cleverly.

What are the Think Tank’s recommendations?

The University should enable the use of AI in research and teaching rather than banning it. What’s important, however, is that—when testing it out for the first time, say—this is done in secure environments, such as in “sandbox systems,” where no data is transferred to any third parties. We also believe it harbors great potential in the administration, where many tasks are process-based: AI can help fill out or check forms, for instance. Given the demographic change that’s under way, with many staff on the verge of retirement, this is a major opportunity that we should be seizing in order to relieve the burden on colleagues. 

What impact is AI having on teaching?

This poses one very specific question, namely how we want to run exams in the future. When ChatGPT was released, I tried out some exercise questions—and was astonished by how well the model was able to solve them even though they were extremely difficult from a mathematical perspective. As far as I’m concerned, this means that, if there was another pandemic and online written exams were back on the horizon, we couldn’t repeat what we did in 2022. It’d no longer fit the times. We’ll need to discuss new curricula, new exam formats and new forms of academic and scientific work. There was general agreement in the Think Tank about that. 

And how is AI changing the actual process of learning?

We’ll no longer just be able to teach students how to write or program something but also, and in particular, how to check their results. This is because AI can hallucinate, meaning that it generates statements that are wrong but that sound plausible. And this is precisely why the ability to tell right from wrong is becoming increasingly important. This affects students, teachers, examiners—every one of us, ultimately.

What role is AI playing in research at the University of Bonn?

A very big one. AI has been an integral part of the Clusters of Excellence for a long time now. In PhenoRob, for instance, we’re developing automated systems for agriculture that apply fertilizer more intelligently, harvest crops more efficiently or irrigate them more effectively. In physics, meanwhile, AI is helping to analyze vast quantities of data, such as images from the James Webb Space Telescope. It’s clear that AI is a must-have for anyone who wants to do top-level research, because many discoveries simply wouldn’t be possible without AI systems.

Do you have a message for those who are feeling unsettled by AI? 

Don’t worry—nobody needs to become an AI expert. Take smartphones, for instance. What they can do would have been unthinkable 20 years ago, but now we use them without even thinking about it. It’ll be a similar story with AI applications, I’m sure of it. This is why it’s important for the University of Bonn to put the right structures in place now so that none of its staff get left behind, with training that everyone can understand and a range of safe and secure applications.