James Fern

YouTube video by Machine Learning Street Talk IS THE MIND REALLY FLAT?

I loved the book, but am having trouble reconciling its main ideas with how Nick discusses AI here, especially the final section on the ‘singularity’. Feels like he’s arguing for some form of human exceptionalism? youtu.be/5cBS6COzLN4?.... I wonder if his thinking has now shifted?

Kendzior HIDING IN PLAIN SIGHT (2020) four pages of text on Trump, Epstein, Michael Cohen, death threats to rape victims

Kendzior HIDING IN PLAIN SIGHT (2020) four pages of text on Trump, Epstein, Michael Cohen, death threats to rape victims

Kendzior HIDING IN PLAIN SIGHT (2020) four pages of text on Trump, Epstein, Michael Cohen, death threats to rape victims

Kendzior HIDING IN PLAIN SIGHT (2020) four pages of text on Trump, Epstein, Michael Cohen, death threats to rape victims

On Trump, Epstein, and their social circle of pedophiles. From HIDING IN PLAIN SIGHT, written in 2019, before Epstein was indicted.

"The story of Jeffrey Epstein is one that will define our era..."

us.macmillan.com/books/978125...

"Scientific literacy has long been a cornerstone of higher education but the open research movement has redefined what it means to be literate as a researcher" with @flavioazevedo.bsky.social & @drcpennington.bsky.social

www.timeshighereducation.com/opinion/if-p...

The real crisis isn’t cheating. It’s mistaking marks for learning. With Brian Stout In recent days, the Australian has painted a stark picture of higher education under siege: empty lecture theatres, students “outsourcing” degrees to generative AI, and universities c...

“A striking feature of recent reporting is how much attention is given to preventing and detecting cheating with AI detectors, bans on AI use and invigilated exams, presented as the “only way” forward. Detection is not assurance, and it never was.”
www.linkedin.com/pulse/real-c...

just finished this today - thoroughly enjoyed, thanks for the nudge! Lots of similarities I thought to 'The Experience Machine' by Andy Clarke, 'Edge of Sentience' by Jonathan Birch and 'Being You' by Anil Seth. All worth a look!

AI may densify learning activities & (quietly) reshape expectations of speed, scope, polish etc. Students might feel more capable & productive with AI, take on more work & experience cognitive overload disguised as 'efficiency'. AI literacy may need to include workload literacy & judgment.

partly this was motivated by the issue of 'perverse incentives'. The sectors seen cases where when presented with '24hr exams' (implemented to give flexibility to choose when to complete an exam, that should take 2 hrs), students spent the entire 24hrs on the exam believing more must be better 🤦‍♂️

really interesting. particularly the suggestions for how 'AI Practice' can help (i.e., pauses, sequencing and grounding). Am interested in how this might translate to the student experience in higher education.

AI Doesn’t Reduce Work—It Intensifies It One of the promises of AI is that it can reduce workloads so employees can focus more on higher-value and more engaging tasks. But according to new research, AI tools don’t reduce work, they consisten...

good article

Yep agreed. Plus a more stable back 4 and a goal keeper who doesn’t throw it in his own net is a huge bonus!

Still not having him 😉

4 sparrows in a tree looking into a window

Working on a lecture, whilst being watched…

Smart Glasses in Education - Artificial intelligence Why look at smart glasses now? Smart glasses have been commercially available in some form for over a decade. More recently, they have returned to popular conversation, driven largely by Meta and its ...

Excellent piece on smart glasses and education - hard not to feel that the sector is far from ready in responding to the widespread adoption of such tech nationalcentreforai.jiscinvolve.org/wp/2026/01/2...

Love United Hate Glazer

#GlazersOut 🟩🟨

2nd trip to OT for the nephews ❤️

Paris said au revoir to cars. Air pollution maps reveal a dramatic change. Air pollution fell substantially as the city restricted car traffic and made way for parks and bike lanes.

“Air pollution fell substantially as Paris restricted car traffic and made way for parks, people-streets and bike-lanes.”

Better for the climate, better for health, better for livability and quality of life.

Common sense.

Such a no-brainer, it’s remarkable that more cities HAVEN’T done the same.

Am generally supportive of a ban (am not sure we need to 'wait for evidence', and a year seems arbitrary), but only if it comes with much stricter regulation for everyone, given its negative effects don't harm just children.

I added this to my ‘To Read’ pile in June but just have not got around to it 🤦‍♂️ thanks for the nudge!

Thanks for sharing, hadn’t seen this but it looks very interesting!

Yes I think I’ve seen something similar with reference to math questions. It gets the right answer, you then ask how it did it and it makes up a process that wouldn’t have got the answer correct. I may be misrepresenting that, sorry!

Back-scratching bovine leads scientists to reassess intelligence of cows Brown Swiss in Austria has been discovered using tools in multiple ways – something only ever seen in humans and chimpanzees

COW TOOLS COW TOOLS COW TOOLS

You may have read them already but I found Andy Clarke (The Experience Machine), Blaise Aguera y Arcas (What is Intelligence?) and Nicholas Humphrey (Sentience) excellent on these topics.

I saw a paper (which I now can’t find!) about the distinction between a models sometimes very certain response when asked about confidence in its answer and the actual ‘under the hood’ reasoning suggesting it wasn’t ‘certain’ at all during inference. Got me thinking about ‘choice blindness’ etc

Oh I agree completely. Do you think it’s helpful to think of LLMs with this framework? E.g., The responses it gives are those of an interpreter that creates a coherent narrative to explain actions (weights?) initiated elsewhere in the brain (neural network?).

Or to @catblanketflower.yuwakisa.com example. The language centre tells ‘us’ the topic must be stressful (post hoc) because another part of the brain registers a change in heart rate, blood flow etc?

I’ve not fully thought any of this through deeply but am drawn to ideas from post hoc rationalisation, choice blindness, split brain patients etc to try to understand modern LLMs I.e., an interpreter that creates a coherent narrative to explain actions initiated elsewhere

Not watched (yet), is the idea that humans have loss functions tied to things like oxygen debt, temp, energy depletion, social exclusion and isolation etc which are ongoing and unavoidable. LLMs optimise loss during training, but at inference no persistent internal variables are at stake?

AI can perform many discourse markers of learning & judgement. HE can't continue to treat these as internal properties that become visible in text outputs. They must be conceptualised as developmental - evidenced through a students sustained pattern of reflections, revisions & actions over time.

If humans can’t distinguish between 1-3 paragraphs of human and AI text, why do you think that would change over longer pieces of writing?

The studies cited aren’t all on 1-3 paragraphs.

Have you used these tools much? They comfortably produce thousands of words of text, fully citing sources, with links.

It’s unlikely a study that blinds markers to mark multiple AI written thesis length documents would get IRB approval.