
Why intellectual virtues matter in AI-driven research
What human qualities do we need to cultivate in the age of AI?
Generative AI is rapidly changing scientific research, offering new possibilities but also raising questions about bias, reliability and independent thinking. At a conference at TU/e on 5 and 6 November, researchers will explore how intellectual virtues can help scientists use AI responsibly.
From searching scientific literature to writing code and analyzing data: generative AI is becoming an increasingly familiar tool in academic research. But as researchers rely more heavily on AI, an important question arises: how do we ensure that we do not simply accept what these systems tell us?
For Vlasta Sikimić, Assistant Professor of Philosophy of Science at TU/e, the answer lies partly in cultivating intellectual virtues: human qualities that help us think, learn and make responsible decisions. These include curiosity, intellectual humility, courage and intellectual justice.
“A virtue is a disposition to act or learn well,” she explains, referring to a philosophical concept that goes back to Aristotle. Rather than relying exclusively on rules and guidelines, Sikimić argues that researchers need to develop these qualities to remain intellectually independent when working with AI.
She illustrates the difference with an example from education. Imagine two students who both provide correct references in an assignment. One does so because the teacher requires it; the other because they genuinely believe it is important. Although the outcome is identical, their motivations are different. The first student might stop checking references when there are no rules requiring it, whereas the second is more likely to maintain the habit.
The same principle applies to AI. Rules can tell researchers what to do in specific situations, but they cannot anticipate every new development. Intellectual virtues, Sikimić argues, help people make responsible decisions even when no specific guidelines exist.
Intellectual justice, humility and openness
One particularly important virtue is intellectual justice: the willingness to engage fairly with knowledge and the people who produce it, such as taking different perspectives seriously and recognizing when certain viewpoints are underrepresented.
AI systems learn from existing data, which means they can reproduce biases already present in society. Sikimić gives the example of researchers using AI to generate hypotheses. If an algorithm primarily draws on research questions relevant to Western societies, other perspectives may be overlooked. Researchers need to recognize these limitations and be willing to challenge the suggestions they receive.
This is also a concern for Yara Daamen, Research and Education Officer at TU/e, who works on AI fairness and ethics. She believes students need more education about the social implications of the technologies they are learning to develop.
“We’re really good at creating these really good engineers that make really good systems, but we don’t think about this bias of what goes behind it,” she says. Technical accuracy alone is not enough if the underlying data or assumptions are biased.
Recognizing deepfakes
Sikimić explores this approach in her NWO-funded project Intellectual Virtues and the Future of Research with Generative AI. She recently published a philosophical framework for responsible AI use in scientific research.
The project also translates these ideas into education. Together with Daamen, she teaches a module on intellectual virtues in the course Critical Thinking in the Age of AI, which attracted 311 students this year.
The course uses real-world cases to help students recognize and question how AI shapes what counts as knowledge. In one case, AI tools used by English councils described the same health needs as less serious for women than for men, which could lead to women receiving less care.
Another looked at the deepfakes an AI-generated fake opposition program spread during Hungary’s 2026 elections, and how flooding the information environment with fabricated content makes it harder for people to form accurate beliefs.
Daamen says students particularly appreciated the latter example. “Students become aware of the risks and start reflecting on how to deal with them. That’s basically how you develop intellectual character.”
From research to practice
The upcoming conference, The Future of Science: How Intellectual Virtues Guide AI-Accelerated Science, brings together the project’s research and educational activities.
During the two days, researchers, students and policymakers will discuss which intellectual qualities are essential to scientific research and how these can be maintained as AI becomes more involved in the research process.
The program includes talks, a panel discussion and specialized workshops. In a workshop led by Daamen, participants will explore how AI affects their work. They will reflect on the intellectual virtues involved, identify areas for improvement and develop concrete actions to make their research practices more responsible.
The panel discussion brings together researchers and policymakers. The organizers also aim to connect people working on these issues and build a broader community around intellectual virtues.
Rather than offering ready-made answers, Sikimić and Daamen want participants to critically reflect on their own practices and exchange ideas.
Ultimately, the challenge is not simply to make science faster with AI, but to ensure that it remains reliable, fair and trustworthy. That, they argue, requires not only better technology, but also researchers who know when to question it.
Participation is free of charge. For more information and to register, visit the website.


Discussion