What students should know about the pros and cons – think about the ease with which ChatGPT can summarise a text or develop an idea, but also the danger of inaccurate or incorrect results – is one thing, said Sjoerd Claessens, associate dean of education, but there is much more at play with AI. Privacy problems, for example. The Dutch Data Protection Authority recently warned about using Deep Seek, a Chinese chatbot, after concerns about its privacy and reliability. SURF, the ICT cooperative of education and research in the Netherlands, had this to say about Microsoft’s alternative, Copilot: ‘Don’t use it’, also due to privacy concerns. What about Open AI’s ChatGPT? That last one is an American company with no interest in European laws and regulations, is the conclusion in the Law Council. And what about aspects such as sustainability? Training and using generative AI tools lead to an enormous amount of CO2 emissions. And another thing: finances. What if, in the future, UM purchases an AI tool that does comply with European guidelines, what would a licence cost? Forty euros per student per month? The council wants to know if the organisation would be willing to pay that.
Brainstorm session
Mark Kawakami, council member on behalf of academic staff, wondered aloud whether individual lecturers are allowed to use ChatGPT, for example, for an assignment in their own teaching groups. “There is a need, a wish,” he emphasised. “And it also has to do with employability. We want to educate students well. If we do nothing, we stand still, and ultimately, someone will suffer as a result.”
Anke Moerland, associate dean of research, understands the urgency. “We have to teach students how to use it. They have to know that you can’t just upload any and every document. So, let’s invest in AI skills.” Teachers also have a part to play, she added. They can explain to students at the start of a course what they can and can’t do. So, not generating a text for a written exam, but maybe for a brainstorm session, for example.
Training
First and foremost, the discussion in the Law Faculty Council is by no means the only discussion about AI at UM recently. The subject has featured in other faculty council meetings, too, such as the Faculty of Arts and Social Sciences, where the need for more oral exams was raised.
While there has been a university-wide ‘policy guideline’ – which according to the description, is a document “which describes UM’s position regarding generative AI and sets out general guidelines” – for some time, it doesn’t offer much advice, the University Council complained a few months ago, when the document was subject to discussion. All the while, the university community is desperately looking for something to hold on to, they warned. How are students, lecturers, but also Boards of Examiners, supposed to tackle urgent AI issues? What should be decided in the faculties and what should be done ‘centrally’?
Inequality
The Executive Board would have liked to develop a uniform AI policy for the whole of UM, but that was seemingly impossible. So instead, it is up to the faculties to work out the exact details. But the U-Council wanted to know whether there will be ‘central’ money and guidance available to safely and responsibly implement AI. And what about training staff and students, who is responsible for that?
Another important point raised, just like in the Law Faculty Council: which AI program will UM choose, because everybody should use the same version to avoid inequality. After all, the paid version of ChatGPT offers more options than the free version. At the time, the Executive Board said that UM was ‘too small’ for offers from tech giants. They were hoping for a collective purchase, so together with other institutions, through ICT organisation SURF, for example. It is unclear whether there has been any progress regarding the purchase of an AI program. At time of publication, the spokesperson for UM had not answered Observant’s question.