INFORMAZIONI SU QUESTO EPISODIO
University and college are where we send young people to get ready for the working world. Whatever’s happening there determines the future of workers, and right now, it’s being redesigned by people improvising in real time.
Dr Sam Illingworth is a full professor of critical AI literacy at Edinburgh Napier University. He writes Slow AI here on Substack, and he wrote a book of the same name. He also has a PhD in atmospheric physics, a decade of published poetry, and a board game about decarbonizing a city.
Sam’s argues that our institutions measure the wrong thing. We built assessment around output because output is what we knew how to mark. AI produces output beautifully, so output is now a measuring instrument that no longer works.
“You cannot tell a student not to use AI and then use AI to write your curricula, grade their papers and edit your feedback.”
That applies to work, too.
What you’ll learn in this episode
* Why assessment has measured output for decades, and what Sam thinks it should measure instead
* What happens to a graduate who was told for three years not to touch the tool their first interviewer asks about
* Why the 9-5 job Sam was promised growing up does not exist any more, and why he does not think that is a bad thing
* Why communication, judgement, critical thinking and empathy are learned skills
* Why smartphones in the lecture hall are probably not doing what you assume
* What Sam tells a parent whose kid is choosing their path at age 14, and what he would no longer say about “study what you love”
* The one thing he would change about universities tomorrow
* How a decade in climate research left him optimistic rather than paralyzed
Chapters
0:00 Ten percent cheat, and why that’s not the story
1:09 The future of workers, not the future of work
2:33 Don’t call him doctor
3:13 Slow AI: when to use it, when to leave it alone
4:17 Poetry, board games and a city at carbon zero
6:12 Why ChatGPT made him pivot everything
8:23 Process over output: the case for pausing
9:14 We assess output. We should assess change.
10:10 “AI cheating is a red herring”
11:42 Don’t tell them no while you use it to grade them
12:29 The graduate who was never allowed to use AI
14:15 The nine to five I was promised doesn’t exist
16:09 Do we still need university?
20:00 Palantir’s founders and what they studied
22:07 Translating between human and machine
23:39 Nobody is born good at this
25:11 The phones in the lecture hall
27:25 Ask the students, they’ll tell you
30:14 Choosing at 14
32:39 The apprenticeship is the way forward
33:40 The magic wand: Put students in the room
36:18 What the Montreal Protocol proves
38:40 Who’s next?
Links and resources
Follow Sam Illingworth:Website - https://www.samillingworth.comBluesky - https://bsky.app/profile/samillingworth.comLinkedIn - https://www.linkedin.com/in/samillingworthSubstack - https://theslowai.substack.comTikTok - https://www.tiktok.com/@theslowai
Referenced in the episode:
Slow AI: Knowing When to Use AI and When to Leave It Alone, Sam Illingworth, self-published, 2026.
Carbon City Zero, the deck-building game Sam co-designed with Paul Wake, published by 10:10 Climate Action, 2020.
The student survey:
Findings are presented in Gow and Illingworth for Advance HE, 9 April 2026. A paper on the survey itself is in preparation and has not been published or peer-reviewed.
* The headline number, in Sam’s own write-up: more than 10% of students admit using generative AI for all or most assessments even when told not to, with a further 20% sometimes. The habitual share is around 1 in 10, and roughly 30% admit breaking the rule at least sometimes.
The MIT report: Final report of the MIT Ad Hoc Committee on AI Use in Teaching, Learning, and Research Training, released 13 August 2026.
Graduate apprenticeships, and the Canadian answer
* Canada has no national degree-apprenticeship system. Ontario’s apprenticeship system covers skilled trades and ends in a Certificate of Qualification, not a degree. Co-ops are paid, but you still have to pay tuition.
* The closest genuine match is York’s Lassonde Digital Technologies BASc, which York describes as Canada’s first fully work-integrated degree: an employer hires you on a renewable full-time contract for up to four years, you earn a salary, roughly 80% on the job and 20% academic, and you finish with a bachelor of applied science. It launched in 2023.
Also named:
* Lev Vygotsky and the “more knowledgeable other”
* Yann LeCun on world models
* Entrepreneurial Campus at Edinburgh Napier
* John Brewton, who writes Operating
Full transcript and resources: [LINK NEEDED: Substack episode post URL, once published]
If this made you think of a parent asking what their kid should study, send it to that parent.
Next weeks on Stacks on Deck: Hannah Busby, who writes Over Capacity, on toxic workplaces and why your job may be making you dumb.
Get full access to Human+AI at nicolleweeks.substack.com/subscribe
