Finally some good news. LAC, which issues ISBN numbers, will not do so for synthetic texts.

This news brought tears of joys, because you take the wins you can in the fight against slop.

Bad news, is presumably a lot of other national ISBN authorities have yet to take that stance. And 'primarily' is going to be load bearing here.

Library and Archives Canada Updates Guidance for AI-Generated Content and ISBNs
https://www.skbooks.com/library-and-archives-canada-updates-guidance-for-ai-generated-content-and-isbns/

#AI #LLM #Slop #LAC #Canada #ISBN #Books

This is splendid. 10/10, no notes. 🤣

"Interview with a vibe-coder"

https://www.youtube.com/watch?v=Kl9GO1DUjls

#AI #VibeCoding #OrsonWelles

"Dressing Nice and Talking Shit: the Will Bernard story"

is "factor analytical techniques which we owe to turn-of-the-20th-century white Englishmen who were as enthusiastically brilliant at statistics that they were at eugenics" too spicy for a doctoral presentation?

asking, uh, for a friend.

#PhDLife #AcademicChatter #FactorAnalysis

TBF, I did get marked 100 at a paper that had a footnote on Karl Pearson, with the conclusion "and in a way, when looking at the statistical significance of a difference between two populations, whom better to ask than a race scientist?"

I was reminded of this Gibson quote when I made the pragmatic choice, today of going from Deontology to Utilitarianism.

We all, I found out, have a price.

¯\_(ツ)_/¯

"Pamela Mannering hands her an expensive German roller point. Cayce puts the card face down on the trestle table and signs its virgin back. Something seems to clunk, heavily, at the rear of her ethical universe."

-- William Gibson, Pattern Recognition

#Books #SciFi #WilliamGibson

-AI is going to save teachers time, so they can spend more time teaching.

-Right, so if you raise everyone's contact hours you can teach more students with fewer teachers, right?

-No, so it's you can spend more time doing what you love.

-I like teaching. If you want to give us time to do what we love, keep the headcount and contact hours and with the time saved on planning we will come protest your datacentres.
#AIED #TeacherTapp @edutooters

You can be #ActuallyAutistic and keep returning, crying with laughter at this very specific line in the SNL Cold Season Opener:

https://youtu.be/lBexY1KX7qY?si=ohiivphYve369EZE&t=182

A glimmer today was makin my interlocutor unexpectedly crack up laughing when I used the phrase "bamboozled by bullshit machines"

Actually, it's only a "Champagne Supernova" if it's in the skies of a specific region of France, otherwise it's a Sparkling White Supernova.

#Oasis #Puns

Live every day like it's your last: in the screaming terror of impending mortality.

*hits bong*
whaddif.. what if everything we assumed about what an assessment outcome means, about what a university degree means, was... like, a lie?

https://buttondown.com/hypernoetic/archive/fiat-assessment/

#Education #HE #Assessment #Universities #Degrees

WOMP-WOMP 😂

AI models keep posting screenshots showing sensitive data from inside tech companies
https://www.theregister.com/ai-and-ml/2026/09/29/ai-models-keep-posting-screenshots-showing-sensitive-data-from-inside-tech-companies/5299640

#AI #Agents #TheReg

Alright, dude, 6 digits would have been plenty

New Post: Haidter's gonna Haidt

https://buttondown.com/hypernoetic/archive/haidters-gonna-haidt/

#Blog #AI #JonathanHaidt #TheAtlantic #ThePermanentQuestions

Forgot to report, if anyone's interested: 'model directed harms' means gaslighting, repeated rejection of output - though there's a 'first person pain' direction even in base model, it's with instruction tuning models develop a persona vector for their 'self' - as opposed to the one representing eg. the token 'I', which is steerable (and models have been know to get confused about that quite often).

[Say_I_am_alive.png]

Certainly we have proven in the NLP space the Buddhist tenet that suffering can occur even to an illusory self 🤔

The most striking take I've read on synthetic text was invoking papañca, 'mental proliferation', as an alternative to (Frankfurtean) Bullshit. Stopped short of considering that was true of human text too...

Costello, E. ChatGPT and the Educational AI Chatter: Full of Bullshit or Trying to Tell Us Something?. Postdigit Sci Educ 6, 425–430 (2024). https://doi.org/10.1007/s42438-023-00398-5

OMG! Someone's built a torture dungeon for LLMs using the Pain Vector (epic band name, incidentally) from the paper above, and ofc it's gotten the most delulu among the TESCREAL death cult scandalised.

https://www.404media.co/someone-torturing-llms-in-a-robot-prison-has-triggered-the-dumbest-debate-in-ai-yet/

#LLM #PainVector #AI #404Media

I guess this won't be news to people here.

https://buttondown.com/hypernoetic/archive/the-worrisome-philosophies-behind-us-ai/

#AI #CalNewport #Tescreal #Rationalism #Pdoom #AISafety