University course explicitly focused on high-level psychology of AI more than humans by end of 2023
31
128
610
resolved Jan 9
Resolved
NO

must be written in the course description in a way verifiable to those outside the university in question. being heavily discussed in an ordinary human psychology class does not count (but would be very cool).

clarification: resolves based on courses that have already occurred, but only ones that study large language models, and only ones that do so using whole-network approaches to analysis. "neuroscience for ML" doesn't count - mech interp of individual circuits is too small; it has to be psychology-scale (at this stage, presumably that means psychology-vagueness; that's okay). It also can't be a machine learning engineering or capability research class, unless it gets deep into analyzing the behavior of learned models at system scale rather than circuit scale.

I will not bet.

Get Ṁ1,000 play money

🏅 Top traders

#NameTotal profit
1Ṁ305
2Ṁ213
3Ṁ160
4Ṁ150
5Ṁ136
Sort by:

Resolved no as no proof was posted in comments before close and I couldn't find any.


@L Can you please resolve?

@traders Have any proof of resolution in case the Creator does not return in due time?

@EdvardsKarlisVonsovics you're a lot more confident than I expected. do you already know of a yes resolution?

bought Ṁ10 of NO

https://thenewcentre.org/archive/artificial-neurosis-ai-psychopathology/ this could fetch you a few masters-level credits when it ran

predicted YES

@ValentinGolev That looks really cool! The new centre looks like a cool org, I've encountered them recently while browsing around interesting stuff with a modern search engine. However, I don't think they're an in-person org? I'm not sure I'm doing a great job researching them.

To clarify explicitly: this will resolve based on courses that have already occurred, but only ones that study large language models, and only ones that do so using whole-network approaches to analysis. "neuroscience for ML" doesn't count - mech interp of individual circuits is too small; it has to be psychology-scale (and, at this stage, presumably psychology-vagueness). It also can't be a machine learning engineering or capability research class, unless it gets deep into analyzing the behavior of learned models at system scale rather than circuit scale.

@L yeah that's why I bought NO, I expected you to say "only real LLMs". it was a fun class and I had a chance to write a very weird essay https://tripleampersand.org/can-machine-lack-lacanian-computation/ but it wasn't technically specific at all, we discussed like, tv shows and fiction about AI

I don't think they're an in-person org?

No but they do give out actual college credits if you need them

bought Ṁ10 of NO

A course by the university, counting for credits for example, and not a student group run course in a university, right?