Publications - Guest articles
Who learns research when AI does the thinking?
16.11.2026
AI writes codes, analyses data and supports the search for new substances. But how do young researchers learn their skills when they can delegate so many tasks? Markus A. Lill, Professor of Computational Pharmacy at the University of Basel, talks about study time, mistakes and the responsibility of universities and companies.
At the scienceindustries annual event, Markus A. Lill outlined in his keynote speech the potential that artificial intelligence offers for substance research and identified its limitations. In the following discussion with Nadine Lumme, Project Manager for Communications and Public Affairs at scienceindustries, the focus turned to another question: What skills must young researchers develop themselves if AI is taking over more and more of their work?
For Lill, the answer starts with the basic principles: if you want to evaluate results, you need to understand how they are produced. That is why early-career researchers should first carry out key steps themselves before automating them.
Doing the whole process yourself once
In his computer-based research, Lill allows young researchers to carry out calculations and simulations independently right from the start: from preparation through calculation to the analysis of the results. When working with data, they should also first understand its structure, quality and significance, develop their own hypotheses and interpret the results themselves.
Furthermore, researchers who develop new methods should be familiar with the underlying algorithms and understand how the code works. This also applies when AI assists with programming: researchers must be able to understand, check and, where necessary, improve the generated code.
This takes time at first. But those who work their way through complex problems at an early stage gain something that cannot be produced overnight: scientific experience. It helps them to tackle more challenging questions independently later on and to assess the quality of their own research.
When the first attempt fails
Lill emphasises how important it is to have the freedom to experiment. Research rarely follows a straight path: approaches fail, results surprise us, and hypotheses turn out to be wrong. It is at such moments that experience is gained. Those who can recognise and correct their own mistakes are also better prepared to evaluate answers provided by AI.
At the same time, Lill encourages young researchers to make use of AI. It enables them to automate repetitive tasks and free up time for scientific inquiry and the interpretation of results. AI must not replace independent thinking, but it can help to clarify relationships, find new perspectives or challenge hypotheses.
Lill recommends first thinking through possible explanations yourself and developing your own hypotheses. Researchers should then systematically check the AI’s suggestions for plausibility and compare them with scientific findings and the specialist literature.
Less content, more collaborative thinking
In Lill’s view, to enable students to develop these skills, universities should place greater emphasis on encouraging critical thinking, creativity and problem-solving. He also advocates reducing the volume of course content where detailed knowledge is not necessary for understanding the fundamental concepts. This would leave students with more time to engage with open-ended questions and work together to find solutions.
He cites the flipped classroom as an example: students work through the basics independently using specified information sources and use the time spent with lecturers to solve problems and discuss their ideas. Examinations could make greater use of presentations, oral discussions or independent problem-solving under supervision to assess whether students understand the interrelationships. However, universities need sufficient staff to deliver such teaching methods and provide individual support – particularly when student numbers are high.
Where will tomorrow’s experience come from?
According to Lill, companies must also ensure that young professionals can gain practical experience. This becomes particularly important as AI increasingly takes over tasks through which those starting their careers have traditionally learnt their skills.
After all, if everyone is looking for experienced professionals, where are they supposed to gain that experience? Lill suggests expanding practical training opportunities for knowledge-based professions, such as industrial placements, structured trainee programmes and joint initiatives between universities and companies.
He advises young people entering the field of research to remain curious, to question established approaches and to try out new methods. For him, this also includes adapting one’s own research work when new findings or tools open up alternative approaches.