I love this blog. It’s well written. But I think the democratisation of cheating is a good thing that’s going to lead to the watershed changes in education that are long overdue. Nothing as slow-moving as education changes without a cathartic crisis.
I’m surprised to find I agree with you Sofia. I am convinced AI can positively transform learning, and schools will resist this until it is impossible to ignore. But I’m also surprised that universities are so disinterested in learning that they ignore AI cheating.
I am surprised I agree with me too! I started out thinking cheating was a problem and have pivoted after months of school inaction and taking up useless wrappers.
Education is iin crisis. The possible futility of many degrees is one, and I wonder how other nation's universities deal with it, or don't? But because students are customers it becomes tricky unless government acts. But also there is a further question about what university is for. I am inclined to the very old fashioned view it's about developing curiosity and a chance to think more deeply and should have nothing to do with training people for work. But I may be dogmatic about this. I think business influence on education and government as a whole has proven to have a negative impact. But that maybe because the people who have been listened to and the people who were listening were not the wisest of choices. I don't know.
I think you're spot on about curiosity. It's a key attitude to develop in learning. I'd take a student with curiosity over and number of A*s any day of the week. Curiosity would be a far better entrance requirement.
A deeper problem is that degrees are mere credentials and most of what you “learn” is utterly worthless for employment. Just cut out the middleman at this point.
I think this is far more the case now than a few decades ago. For those of us going on to H.E. in the latter part of the 20th century going away was also about finding your feet. Many teenagers then lived in authoritarian households, particularly girls. Many parents would countenance bringing a boy/ girl friend home never mind staying over so going away as a student gave you the freedom that seems to be the norm now for 21st century teenagers. I think there was also less certainly about what you might do when you graduated, unless you were studying something like medicine. Putting all the issues linked to course fees to one side, I didn't know anyone who'd calculated which job would offer the best salary. And the teaching and learning then was mostly through tutorials and seminars. I was never in a lecture room with more than 30 other students. I don't think all of my tutors were the bees' knees by any means but I do believe that most of the students I knew across disciplines, were doing degrees because they were genuinely interested in the subject.
Cheating has always existed in most spheres of life. I can certainly remember kids copying other kids homework on the bus. And I 'borrowed' some one's essay at university although back in the dark ages of the 20th century I did have to copy it out by hand and add a few bits and bobs so it didn't look to like the original. I hope the spectre of AI doesn't mean yet more formal written assessments whether at school, F.E. or H.E. What's really needed are assessments that require the learner to demonstrate application of learned skills or knowledge. I did drama and English for my degree and a significant part of the drama assessment was practical. I'd like to see far more of this including at school level. There are plenty of different ways to create opportunities for learners to do this. Much harder to grade of course but grades are pretty pointless really. You don't get a grade when you take a driving test ; you pass or you fail. Getting firsts in your degree doesn't necessarily mean you'll be the best when you apply what you've done in a work situation.
It brings back the oral Q&A exam of old. That was 'the gold standard' when I attended uni. (decades ago) It fell by the wayside b/c it simply doesn't scale, it's an impossible physical burden for the faculty staff. I had witnessed few heroic Prof-s that did it, for the love of teaching, but it was clear, most normal people can’t do it.
It's incomprehensible to me how what’s reported in the link is not an existing product. Why no demand from the education industry, why no supply from the AI industry.
I've been thinking deeply about this topic, and while I agree with your overall conclusion that LLMs are an existential threat for the current model of university education, I think you underestimated the diversity and heterogeneity of responses from universities. Some are putting their heads in the sand, as you implied, but others are working proactively to explore the post-LLM model of university education. Another perspective you missed is that not all academic disciplines are actually threatened by LLM; some actually view it as a transformative opportunity. The disciplines that teach their students how to critically evaluate evidence and have independent thoughts will likely thrive, whereas the disciplines that primarily act as gatekeepers and teach their students what to think will not.
Great stuff. Beyond inertia or the fear of putting off prospective customers, do you have a theory for why more universities haven’t trialled in person assessment? I just think it’s so odd we haven’t seen at least some innovative piloting by a leading university - for PR value if nothing else!!
I think part of the issue is that a lot of them are quite decentralised when it comes to assessment. It’s seen as an aspect of academic freedom. So quite hard for someone central to say “this is how it will work for everyone”
There’s also the mental health aspect. Many students have anxiety which is triggered by the exam hall setting, so that there is little appetite for change from the student body. Furthermore, the complexity of access arrangements generates a huge administrative and logistical load for in-person exams.
Which begs the question - why do you work for an Ed tech startup proactively advocating for the inclusion of AI in teaching? My best guess is that as Von Neumann said of Oppenheimer, “sometimes one confesses a sin in order to take credit for it”.
The long term consequences of AI slop are not terrible because they are not terminal. Try seeing the digital, binary, alphabet and related arbitrary imagery as very dated technology that simply combined into automation made them all irrelevant.
Simply time to innovate what we should have innovated centuries ago,
I love this blog. It’s well written. But I think the democratisation of cheating is a good thing that’s going to lead to the watershed changes in education that are long overdue. Nothing as slow-moving as education changes without a cathartic crisis.
I’m surprised to find I agree with you Sofia. I am convinced AI can positively transform learning, and schools will resist this until it is impossible to ignore. But I’m also surprised that universities are so disinterested in learning that they ignore AI cheating.
I am surprised I agree with me too! I started out thinking cheating was a problem and have pivoted after months of school inaction and taking up useless wrappers.
Education is iin crisis. The possible futility of many degrees is one, and I wonder how other nation's universities deal with it, or don't? But because students are customers it becomes tricky unless government acts. But also there is a further question about what university is for. I am inclined to the very old fashioned view it's about developing curiosity and a chance to think more deeply and should have nothing to do with training people for work. But I may be dogmatic about this. I think business influence on education and government as a whole has proven to have a negative impact. But that maybe because the people who have been listened to and the people who were listening were not the wisest of choices. I don't know.
I think you're spot on about curiosity. It's a key attitude to develop in learning. I'd take a student with curiosity over and number of A*s any day of the week. Curiosity would be a far better entrance requirement.
A deeper problem is that degrees are mere credentials and most of what you “learn” is utterly worthless for employment. Just cut out the middleman at this point.
I think this is far more the case now than a few decades ago. For those of us going on to H.E. in the latter part of the 20th century going away was also about finding your feet. Many teenagers then lived in authoritarian households, particularly girls. Many parents would countenance bringing a boy/ girl friend home never mind staying over so going away as a student gave you the freedom that seems to be the norm now for 21st century teenagers. I think there was also less certainly about what you might do when you graduated, unless you were studying something like medicine. Putting all the issues linked to course fees to one side, I didn't know anyone who'd calculated which job would offer the best salary. And the teaching and learning then was mostly through tutorials and seminars. I was never in a lecture room with more than 30 other students. I don't think all of my tutors were the bees' knees by any means but I do believe that most of the students I knew across disciplines, were doing degrees because they were genuinely interested in the subject.
Cheating has always existed in most spheres of life. I can certainly remember kids copying other kids homework on the bus. And I 'borrowed' some one's essay at university although back in the dark ages of the 20th century I did have to copy it out by hand and add a few bits and bobs so it didn't look to like the original. I hope the spectre of AI doesn't mean yet more formal written assessments whether at school, F.E. or H.E. What's really needed are assessments that require the learner to demonstrate application of learned skills or knowledge. I did drama and English for my degree and a significant part of the drama assessment was practical. I'd like to see far more of this including at school level. There are plenty of different ways to create opportunities for learners to do this. Much harder to grade of course but grades are pretty pointless really. You don't get a grade when you take a driving test ; you pass or you fail. Getting firsts in your degree doesn't necessarily mean you'll be the best when you apply what you've done in a work situation.
Have you seen the “fight fire with fire” done on a shoe string budget?
https://www.behind-the-enemy-lines.com/2025/12/fighting-fire-with-fire-scalable-oral.html
It brings back the oral Q&A exam of old. That was 'the gold standard' when I attended uni. (decades ago) It fell by the wayside b/c it simply doesn't scale, it's an impossible physical burden for the faculty staff. I had witnessed few heroic Prof-s that did it, for the love of teaching, but it was clear, most normal people can’t do it.
It's incomprehensible to me how what’s reported in the link is not an existing product. Why no demand from the education industry, why no supply from the AI industry.
I've been thinking deeply about this topic, and while I agree with your overall conclusion that LLMs are an existential threat for the current model of university education, I think you underestimated the diversity and heterogeneity of responses from universities. Some are putting their heads in the sand, as you implied, but others are working proactively to explore the post-LLM model of university education. Another perspective you missed is that not all academic disciplines are actually threatened by LLM; some actually view it as a transformative opportunity. The disciplines that teach their students how to critically evaluate evidence and have independent thoughts will likely thrive, whereas the disciplines that primarily act as gatekeepers and teach their students what to think will not.
Great stuff. Beyond inertia or the fear of putting off prospective customers, do you have a theory for why more universities haven’t trialled in person assessment? I just think it’s so odd we haven’t seen at least some innovative piloting by a leading university - for PR value if nothing else!!
I think part of the issue is that a lot of them are quite decentralised when it comes to assessment. It’s seen as an aspect of academic freedom. So quite hard for someone central to say “this is how it will work for everyone”
There’s also the mental health aspect. Many students have anxiety which is triggered by the exam hall setting, so that there is little appetite for change from the student body. Furthermore, the complexity of access arrangements generates a huge administrative and logistical load for in-person exams.
A useful and piquant blog; some terrific comments too. I hope you'll forgive me if I respond via a blog post of my own:
https://pvswrites.wordpress.com/2026/03/16/a-provisional-solution-for-the-use-of-ai-in-university-assessment/
Which begs the question - why do you work for an Ed tech startup proactively advocating for the inclusion of AI in teaching? My best guess is that as Von Neumann said of Oppenheimer, “sometimes one confesses a sin in order to take credit for it”.
The long term consequences of AI slop are not terrible because they are not terminal. Try seeing the digital, binary, alphabet and related arbitrary imagery as very dated technology that simply combined into automation made them all irrelevant.
Simply time to innovate what we should have innovated centuries ago,