tweety fish

well, lo, people have built a benchmark using a task called multi-armed bandit, where success depends on iteratively grasping the relative odds of payouts from different choices, and LLMs are (generally) shit at it arxiv.org/pdf/2403.15371 /

arxiv.org

this, posted elsewhere in these sprawling threads, is very interesting on the topic of what emotional representations exist in frontier models and what they might mean: www.thetransmitter.org/emotion/what...

What can AI teach us about ‘emotions’? thetransmitter.org

[ veering wildly off on a tangent ] so actually the strong "but big front" reaction is the kind of intuitive physics reasoning people are really good at but ML/"AI" to this point is pretty hopeless at, to the point that it's one of the biggest remaining gaps with self-driving …

A Survey on Machine Learning Approaches for Modelling Intuitive Physics arxiv.org

as promised here is the paper comparing LLMs to human results; I haven't gone through it carefully enough to vouch for the methodology and for various reasons I have my suspicions about how robust it'll be but it's certainly conceptually interesting: arxiv.org/pdf/2505.09901

arxiv.org

quoted sentence paraphrased from today's Matt Levine which links www.bloomberg.com/news/article... and robinhood.com/us/en/newsro...

Robinhood Launches AI Stock Trading, Purchases on Credit Cards bloomberg.com

Last night it occured to me to wond er if LLMs were any good at gambling tasks. This is important not because it'd be funny for LLMs to gamble but because gambling tasks get used to measure human decision-making under risk /

my new opinion is we should ban the sale of anything that currently gets called 'AI' and also ban use of the term specifically so I shut up about it

All the "will 'AI' replace the humanities" shit is so boring, why aren't we trying to supplant random academic fields with other useful coding tools? Metasploit is going to replace comp lit. We'll be doing historical musicology with linters. Let's go

the story about data center investment vastly outstripping investment in public transportation is a story about capital markets and government and doesn't actually have anything to do with 'AI' as such, people get that, yeah?

You know what would be a fun thing to look at is the correlation between “AI will take all the jobs GET READY”-style marketing and “TAKE jobs no no it’s a great productivity tool” as it became clear to all that it was hopeless at female-coded empathetic/creative work and good at male-coded sw eng

listen, I try not to prejudge this stuff too hard because I think it's important for me, at least, to have my skepticism rooted in up-to-date knowledge but "Robinhood offers in-app agentic trading to its users" is one hell of an alarming sentence

Man, the ridiculous bullshit that people will believe about the ‘AI’ industry bums me out, it’s like the left-coded version of the right’s “all cities are dangerous hellholes”. Resist, resist by all means, but don’t let yourself be grifted insensate by charlatans and comforting lies.

I took that "what kind of AI whatever are you" quiz that'd been going around yesterday and am still mad. obviously every internet quiz is essentially a fair midway fortune teller at best but GOD the questions were horribly written. which, again, is all of these quizzes BUT /