Universities Grapple With AI Revolution, Rethinking Teaching and Assessment

The integration of large language models and generative AI into academia is creating a crisis and revolutionizing how universities operate. The ease with which students can now generate essays and solve assignments without deep engagement with the material renders traditional assessment methods obsolete. In response, universities in Israel and globally are fundamentally altering their long-standing practices. Some are reverting to handwritten exams and mandatory attendance, while others are encouraging the integration of AI into coursework to achieve more impressive results.

Currently, there is no uniform policy across universities, and their leaders admit to being in an "experimentation phase," uncertain about definitive solutions. A new survey indicates that 57% of American college students use AI in their academic work at least weekly, with over a fifth using it daily. A similar Israeli survey shows 58% of students use AI tools daily, but 60% report a lack of clear university guidelines.

Lecturers are exploring various strategies to counter AI misuse, including using detection tools and requiring students to document their writing process. However, the consensus is that AI capabilities will continue to advance, making it difficult to stay ahead. This has led some institutions to "go backward," returning to traditional exam formats that were gradually phased out in favor of assignments designed to foster independent learning and critical thinking.

Princeton University, for example, has ended a 133-year-old "honor code" tradition of unsupervised exams due to AI use. Reports from Harvard and other institutions mention a return to in-person exams, bans on laptops in classrooms, and even a shift back to handwritten assignments. Some American lecturers have resorted to embedding random words in assignments to confuse AI models and expose students who haven't read instructions. Hebrew University's Rector, Oren Shagrir, noted the difficulty of moving away from exams and the current need to return to them, while also encouraging faculty to explore alternative methods like in-class presentations and one-on-one discussions.

Bar-Ilan University is adopting a different approach, requiring students to declare their AI usage and use it ethically. The university has formed working groups to share experiences and develop guidelines. Some law faculty are having students write papers and then briefly reconstruct their main arguments orally without notes. However, university leaders acknowledge that these are still experimental measures, and no fundamentally different solutions have emerged globally.

Despite the challenges, AI is also seen as an opportunity. It can assist students with learning difficulties by providing explanations, and universities are exploring its use to accelerate research. Bar-Ilan is reviewing and updating its undergraduate curricula to align with the changing reality. Technion's Danny Raz emphasized that AI is a central focus, requiring rapid adaptation to educate the next generation of researchers and engineers effectively, noting that final projects in computer science have become more ambitious with AI assistance.