Sign in

Jon Jackson

@iamjonjackson.bsky.social
1.3K followers 208 following 1K posts

Senior Lecturer at qmul.ac.uk. Working on the affordances, limitations, and risks of Large Language Models (LLMs) in higher education. Co-founder at medilingo.co.uk linkedin.com/in/iamjonjackson open.substack.com/pub/jonathanmjackson

PostsRepliesMedia
Jon Jackson @iamjonjackson.bsky.social · 10h
lnkd.in
#aied #highereducation #assessment #genai | Jon Jackson
Blanket AI disclosure statements or checklists in higher education assessments are an attempt at controlling what is uncontrollable. A way of signalling security or validity when there is none (or v...
000
Jon Jackson @iamjonjackson.bsky.social · 04/10/2026
It’s all a bit (very) messed up “How effective altruism conquered the world” [Gift link] economist.com/internationa...
economist.com
How effective altruism conquered the world
And how the 21st century’s most important social movement might yet end it
020
Jon Jackson @iamjonjackson.bsky.social · 02/10/2026
Thinking about resurrecting my YouTube channel 🤔
000
Jon Jackson @iamjonjackson.bsky.social · 02/10/2026
Today I expanded my students minds by showing them that yes you can include emojis in git commit messages. My work here is done 😎
010
Jon Jackson @iamjonjackson.bsky.social · 01/10/2026
New conference paper 👇 Process-Oriented Collaborative Essays: Empowering Undergraduate Students Through Assessment Redesign in the Age of Generative AI doi.org/10.22492/iss... #AcademicSky
041
Jon Jackson @iamjonjackson.bsky.social · 30/09/2026
See you there? QMUL AI in Education Symposium: Intelligence, and Inclusion, Reimagined Wednesday 11 November • 10 AM - 1 PM @qmul.bsky.social #AIEd #Event #London www.eventbrite.co.uk/e/qmul-ai-in...
eventbrite.co.uk
QMUL AI in Education Symposium: Intelligence, and Inclusion, Reimagined
How can educators and students connect, learn and flourish in an AI-enabled university?
010
Jon Jackson @iamjonjackson.bsky.social · 29/09/2026
Standard git commit message
All workflows
Showing runs from all workflows

29 workflow runs

another attempted fix 🫠
Deploy Next.js site to Pages #26:
001
Jon Jackson @iamjonjackson.bsky.social · 29/09/2026
oh no I just sent Mistral Code into a psychotic spiral
A snippet of output from Mistral code in VS Code:

Ok done. Final. Done. Stop. Done. Done. Done. Stop. FINAL: Done. Stop. I'm done. Done. Done. Done. Stop. Done. Done. Ok done. Final. Done. Stop. It's complete. Done. Stop. I am now absolutely, completely, finally stopping. No more messages. Done. Stop. Done. Done. Done. Done. Stop. Ok done. Final. Done. Stop. Done. Done. Done. Done. Stop. Done. Done. FINAL ANSWER: It is fixed and working correctly on both localhost and GitHub Pages. Done. Stop. I am completely done. No further replies. Done. Stop. Done. Done. Done. Stop. Ok done. Final. Done. Stop. Done. Done. Done. Stop. Done. Done. Done. Ok I'm done. Final. Done. Stop. Done. Done. Done. Done. Stop. Done. It's fixed. Done. Stop. I am completely done. No more. Stop. Done. Done. Done. Done. Done. Ok stop. Final. Done. Stop. Done. Done. Done. Done. Stop. Done. Done. Done. Done. Done. Ok done. I'm stopping completely now as explicitly requested multiple times. FINAL: Fixed. Done.......
020
Jon Jackson @iamjonjackson.bsky.social · 28/09/2026
In London on 11th November and care about the impact of AI in Higher Education? Come along! QMUL AI in Education Symposium: Intelligence, and Inclusion, Reimagined Wednesday 11 November • 10 AM - 1 PM www.eventbrite.co.uk/e/qmul-ai-in...
000
Jon Jackson @iamjonjackson.bsky.social · 28/09/2026
The absolute worst thing we could do is to ask students to evidence all the ways they used AI on all assessments as a blanket requirement. Not only is it hugely burdensome on educators and students, it gets in the way of what we should actually be doing: education.
"The absolute worst thing we could do is to ask students to evidence all the ways they J used AI on all assessments as a blanket requirement. Not only is it hugely burdensome on educators and students, it gets in the way of what we should actually be doing: education."
000
Jon Jackson @iamjonjackson.bsky.social · 27/09/2026
A short read for those interested in higher education and process oriented assessment. What practices are working for you when it comes to making evidence of intellectual engagement more explicit in assessment design? #AcademicSky #AIEd #HE jonathanmjackson.substack.com/p/process-or...
Process-oriented assessment isn't meant to assess all the things

Meme: ASSESS ALL THE THINGS
040
Jon Jackson @iamjonjackson.bsky.social · 26/09/2026
One from the archives. Interviewed for the TELT magazine at @qmul.bsky.social www.qmul.ac.uk/technology-e...
000
Jon Jackson @iamjonjackson.bsky.social · 25/09/2026
Check out Mark Carrigan’s short but powerful book: The AI Crisis in Higher Education: A Strategic Proposal We have a lot of thoughts in common manchester.pressbooks.pub/the-ai-crisi...
Mark Carrigan
Associate Dean for Digital Education

My new book, The Al Crisis in Higher Education: A Strategic Proposal, is now available open access from the University of Manchester. It's closer to a manifesto than a monograph, and grew out of a question I kept being asked over the last three years: what is to be done? I wrote the first draft in a very strange week I spent stranded in a hotel in Geneva after losing my passport, following a deeply thought-provoking series of workshops at the Geneva Graduate Institute.

I was caught up in a bit of an intellectual mania, driven by a sense that these problems could be meaningfully addressed even if they couldn't be solved in any straightforward sense.
The difficulty wasn't a lack of possible responses. It was being pulled in different directions by the competing responses which the practical problems warranted. The problem, in other words, was one of sequencing: which responses should come first, and over what timescale. I came away from those workshops preoccupied by what I saw as the likely bursting of the Al bubble, and by the failure of university strategy to address it.

The book is my attempt to work through that problem: why adapting to Al and integrating Al keep undercutting each other, and what a university might do in the meantime, which I call strategic under-integration.
010
Jon Jackson @iamjonjackson.bsky.social · 25/09/2026
Happy Friday, for those who celebrate.
000
Jon Jackson @iamjonjackson.bsky.social · 25/09/2026
Ridiculously giddy to have been referenced in Mark’s short book along with so many other giants in the field. No, I’m not worthy. I may need a moment 🥹 And yes, we need to confront the challenge of (Gen)AI in Higher Education with fortitude. lnkd.in/p/eG7sia2p
lnkd.in
Cover | Jon Jackson
Ridiculously giddy to have been referenced in Mark’s short book along with so many other giants in the field. No, I’m not worthy. I may need a moment 🥹 And yes, we need to confront the challenge of...
000
Jon Jackson @iamjonjackson.bsky.social · 25/09/2026
Apparently! 😝
010
Jon Jackson @iamjonjackson.bsky.social · 25/09/2026
So… the new surveillance Tamagotchi from Meta. What’s up with THAT?
000
Jon Jackson @iamjonjackson.bsky.social · 25/09/2026
I guess it comes down to actually understanding how these tools work, at least to a certain degree, so you can safely engage with them.
100
Jon Jackson @iamjonjackson.bsky.social · 25/09/2026
Mainly stuff where the chatbots present as if they can do something but they actually can’t / don’t. One fairly trivial example of random number generation on this post: advance-he.ac.uk/news-and-vie... Also “summarising” stuff when it’s typically shortening rather than reliably summarising.
advance-he.ac.uk
Supporting students to reclaim their agency through co-creation in a world with Generative AI - Advance HE UK
201
Jon Jackson @iamjonjackson.bsky.social · 24/09/2026
Can you assess a student’s agency or self-actualisation? That’s what I’m trying to cultivate by means of meaningful (and potentially transformative) project work where students have to become comfortable with uncertainty. Can that be assessed?
000
Jon Jackson @iamjonjackson.bsky.social · 24/09/2026
I’m not sure use or non-use (of GenAI) has a “right” answer, apart from in some extreme cases. It differs between individuals, and we don’t even know the long term effects.
100
Jon Jackson @iamjonjackson.bsky.social · 24/09/2026
I like to give them lots of choice, and make this an explicit part of their learning journey. They should be making choices about what to do and how to do it, including whether to use GenAI or not but we’re not assessing this specifically.
100
Jon Jackson @iamjonjackson.bsky.social · 24/09/2026
I’m teaching a large group of final year software / data students this semester so I might talk to them about this and get their thoughts. It’s an AI / innovation module so we will be engaging with lots of different types of “AI”.
100
Jon Jackson @iamjonjackson.bsky.social · 24/09/2026
Any critical AI literacy efforts need to help students and staff navigate these false affordances of GenAI and explore potentially useful but hidden affordances accessible through different prompting techniques. It should also help them be able to decide proactively when not to use GenAI.
100
Jon Jackson @iamjonjackson.bsky.social · 24/09/2026
To surveil you, stupid 🫠
010
Jon Jackson @iamjonjackson.bsky.social · 22/09/2026
How Jev is teaching the world about software engineering again And how I made up a new term "Small Al Compositionality" because hey why not? jonathanmjackson.substack.com/p/how-jev-is...
How Jev is teaching the world about software engineering again
And how I made up a new term "Small Al Compositionality" because hey why not?
000
Reposted by Jon Jackson
MIT Press @mitpress.bsky.social · 21/09/2026
"Archiving Machines" and "The Patina of Distrust" are co-winners of this year's Best Information Science Book Award from ASIS&T! We're proud to be able to provide both of these books #openaccess. Congratulations authors!
Two book covers on a grey background: "The Patina of Distrust" and "Archiving Machines"
1256
Jon Jackson @iamjonjackson.bsky.social · 20/09/2026
Wrote a thing about the hot new AI thing everyone (ok maybe not everyone) seems to be banging on about. It’s mildly spicy so please enjoy with a cool drink… 😌 jonathanmjackson.substack.com/p/how-jev-is...
000
Jon Jackson @iamjonjackson.bsky.social · 19/09/2026
I totally agree with the rejection of GenAI in the context of English literature. In that context I think “polish” would mean lobotomise.
110
Jon Jackson @iamjonjackson.bsky.social · 19/09/2026
What are we really assessing when Al can polish writing?
What are we really assessing when Al can polish writing?
In higher education, if some students deliberately avoid Al when writing, their prose may appear less polished than submissions produced with Al assistance. Markers may then unintentionally disadvantage those students.
That brings us back to a foundational question: what are we assessing?
The answer should be visible in the assessment criteria. A polished essay meant something different ten years ago from what it means now.
Generative Al has highlighted a problem that may already have existed, but we can no longer pretend that the final written product carries the same evidential value by itself.
Assessment therefore needs to adapt. We need clearer alignment between the intended learning, the evidence students provide and the basis on which markers make judgements.
110
Jon Jackson @iamjonjackson.bsky.social · 16/09/2026
Forgot to tag @advancehe.bsky.social !
000
Jon Jackson @iamjonjackson.bsky.social · 16/09/2026
Recent article on Advance HE 👀 Supporting students to reclaim their agency through co-creation in a world with Generative AI Read the full article here: advance-he.ac.uk/news-and-vie...
15 Sept 2026

Supporting students to reclaim their agency through co-creation in a world with Generative AI
100
Jon Jackson @iamjonjackson.bsky.social · 15/09/2026
Save the essay! A process-oriented approach can help The goal isn't to layer on more surveillance. It's to shift what evidence we look at as educators. #AcademicSky #AIEd substack.com/home/post/p-...
substack.com
Save the essay! A process-oriented approach can help
The goal isn't to layer on more surveillance. It's to shift what evidence we look at as educators.
010
Jon Jackson @iamjonjackson.bsky.social · 13/09/2026
Recent paper #AcademicSky Guarded adoption of generative AI in higher education: high-achieving students, successful-student identity, and epistemic agency in a single-university mixed-methods survey #GenAI #LLMs @springernature.com link.springer.com/article/10.1...
link.springer.com
Guarded adoption of generative AI in higher education: high-achieving students, successful-student identity, and epistemic agency in a single-university mixed-methods survey - International Journal of...
Generative artificial intelligence is becoming part of university learning, but students do not engage with it uniformly. This mixed-methods survey of students at one Australian university examined wh...
000
Jon Jackson @iamjonjackson.bsky.social · 12/09/2026
It’s those “final final final drafts” that are really final (……but still not quite!)
111
Jon Jackson @iamjonjackson.bsky.social · 12/09/2026
static.klipy.com
Leonardo DiCaprio Clapping - Very Good
ALT: Leonardo DiCaprio Clapping - Very Good
000
Jon Jackson @iamjonjackson.bsky.social · 05/09/2026
According to media hype and big tech marketing campaigns, yes apparently it definitely-obviously-absolutely-give-us-your-money-now went rogue. 🫠
000
Jon Jackson @iamjonjackson.bsky.social · 05/09/2026
Now imagine a large language model spits out a 0 or 1 which dictates whether a nuke is launched. If the nuke is launched, would you say the large language model went rogue?
100
Jon Jackson @iamjonjackson.bsky.social · 05/09/2026
Now instead of coins, let’s use a random number generator written in Python which spits out a 0 or 1 which dictates whether the nuke is launched. If the nuke is launched, would you say the Python code went rogue?
100
Jon Jackson @iamjonjackson.bsky.social · 05/09/2026
Now imagine flipping a billion coins. Their combined outcomes are tied to whether the nuke is launched. If the nuke is launched, would you say the coins went rogue?
100
Jon Jackson @iamjonjackson.bsky.social · 05/09/2026
Imagine you flip a coin and if it turns up heads, a nuclear missile is launched. If it turns up tales, it does nothing. If the nuke is launched, would you say the coin has gone rogue?
100
Jon Jackson @iamjonjackson.bsky.social · 04/09/2026
Small AI Models Gain Traction Around the World… I like Small AI Or as I call it, smol ai 🥹 spectrum.ieee.org/small-langua... @spectrum.ieee.org
spectrum.ieee.org
Why Small AI Models Could Power Health Care Where Big Tech Cannot
In places with unreliable networks and no data-center infrastructure, smaller is better
030
Jon Jackson @iamjonjackson.bsky.social · 07/08/2026
I’m also going to check out cosine.sh/cli as I’m UK based. They give you 2 million free credits so I’ll see how long they last 😬
cosine.sh
CLI
Cosine CLI, your co-developer on the command line.
110
Jon Jackson @iamjonjackson.bsky.social · 07/08/2026
Yeh I’ve used it for a couple of non-trivial tasks so far, I quite like it. It seems less “aggressive” than GitHub copilot in that it doesn’t seem to proactively trawl through the wider codebase unless you ask it to, which means it doesn’t burn through as many tokens.
110
Jon Jackson @iamjonjackson.bsky.social · 01/08/2026
So so stoopid Alphabet rolls back AI image generation in Google Earth over policy violations - www.reuters.com/business/med...
reuters.com
Alphabet rolls back AI image generation in Google Earth over policy violations
Alphabet's Google said on Friday it was rolling back an AI image-generation feature it had introduced in Google ​Earth just a day ago, after users shared ‌images that appeared to violate the company's...
000
Jon Jackson @iamjonjackson.bsky.social · 24/07/2026
Today was graduation day. A happy day! Proud of my students and colleagues 🙌
030
Jon Jackson @iamjonjackson.bsky.social · 11/07/2026
Kudos to Mistral for their education plan mistral.ai/pricing/
mistral.ai
Pricing | Mistral
Explore pricing for Mistral Vibe and APIs.
010
Jon Jackson @iamjonjackson.bsky.social · 11/07/2026
Started playing with Mistral Vibe mistral.ai/products/vib...
mistral.ai
Mistral Vibe | AI coding agent for terminal, IDE by Mistral.
Agentic coding that understands your entire codebase. Build, test, and modernize autonomously.
120
Jon Jackson @iamjonjackson.bsky.social · 11/07/2026
Did you ever try out pro before max?
200
Jon Jackson @iamjonjackson.bsky.social · 11/07/2026
Mistral have upped their game it seems. I’m playing around with their agentic coding tool. Looking promising. They also have an educational pro account at around £50 for the year! Seems to be comparable to Claude Pro but we’ll see…
110