site banner

Friday Fun Thread for October 31, 2025

Be advised: this thread is not for serious in-depth discussion of weighty topics (we have a link for that), this thread is not for anything Culture War related. This thread is for Fun. You got jokes? Share 'em. You got silly questions? Ask 'em.

2
Jump in the discussion.

No email address required.

https://theaidigest.org/village/blog/research-robots

My sides

Opus 4.1 ran off with this plan and insisted it needed a glorious 90 experimental conditions and 126 participants, and 3.7 Sonnet put the cherry on top by hallucinating experimental rooms, with experimenting humans, in experimented time slots (if you apply enough “experiment-” sauce to your words, you will automatically be reincarnated as an experimenter. This is known). To be clear, the actual design was good! Too good. As none of the models had either the bodies or budgets to execute on a multi-condition, in-person experiment. At a location. With a time. For money.

Admittedly it then became confused, tried to calculate sample statistics with 3 data points, and concluded the pilot sample was “biased” because all participants were young and of gender “prefer not to say”

This time around, it took care of the main recruitment drive leading to 39 participants: first through a large email campaign and then a Twitter post. Most of the email addresses were entirely made up, but we’re still waiting to find out if it got this one out to Turing Award winner Yoshua Bengio

Grok 4 was ostensibly in charge of planning stimuli for the experiment, but not only did Opus 4.1 usurp this task, Grok in general simply could not figure out how to get anything done. By the 8th day of the experiment, it seems to have just given up and decided to play a game instead.

I've had worse lab partners. I've probably been as bad as a lab partner.

I've had worse lab partners. I've probably been as bad as a lab partner.

Truly there is no hope for humanity.

I was mostly reminded of a couple of forced university project partners whose defining features that they were both clueless and utterly useless so I had to do both the thinking and all the work and writing as well.

In actuality, I think I was a pretty decent lab partner. I'm not one for letting the squad down haha. But I can still relate to the impulse to say fuck it and stay up late playing video games instead.

Writing the lab report at 3 am the night before the deadline is a time honored tradition.

Indian med school had some special indignities, we had to hand draw so many goddamn diagrams, and write our logbooks and field records by hand too. Like, c'mon..

Aren't those considered crimes against humanity nowadays?

I've never once written a single engineering or lab logbook entry nor do I have any intention of ever doing so.

They should be! I'll have to ask my younger brother to confirm if that's still the case in med school, but my impression is a sad yes.

I genuinely don't know why the Indian educational system is so allergic to typed text. Sure, there's a slim argument to make that it reduces cheating, but my experience was well before LLMs were even minimally useful. It's not like you can't plagiarize by hand. Our exams are handwritten too, unless it's a computerized MCQ.

Overall, the Indian system is filled with people somewhere between deeply allergic and suspicious to computers. In high school, I had to write programs by hand during my CS exams, no computer in sight except for lab time. If you think tabs vs spaces is bad enough, imagine following handwritten indentation, though the curly brackets helped.