Think you slightly messed up the link

Doesn’t he have a kid to raise or something (…yeah…)? Small miracle he still has custody

Not only are they grade-school level awful, he totally misses the point of the original

https://gwern.net/doc/ai/nn/diffusion/imagen/2026-02-13-gwern-chatgpt-gemini3propreview-kimik2thinking-meme-tomgauldourblessedhomelandtheirbarbarouswastes-toolsforthoughtzettlekastenvsprocrastination.jpg

Why does the kitchen in the "zen of instant coffee" strip have two stoves?

Also: needing Claude to make a pie chart. No, that is someone with a cognitive surrender fetish. I refuse to engage.

My God, it's like the art becomes more bafflingly poor the closer he gets to stumbling into a decent joke. There's a finite amount of quality that has to be balanced between the two, and it's capped at "shit".

I did a shallow dive on the subject of that, he was thrown out of lighthaven primarily because he didn't pay (they let non payers hang out for a bit but not after 11pm) but secondarily because he made people feel uncomfortable. He was not thrown out for probably being a rapist, in fact people were starstruck to meet the author of the rapeblog. His writing is utterly disgusting but the lighthaven account does provide an interesting window into rat spaces that rats themselves won't.

It is also an incredible strength that they actually practice what they preach, and don't exclude someone based on hearsay and "probably a rapist".

🫪

The first of the aforelinked blog posts begins:

When I finished HPMOR, I immediately knew it was the best novel I had read in more than a decade.

Goddamn it, people

It continues

the HPMOR comic

No no no no

Yeah lesswrong is weird about China. Its an ongoing thing. It alternates between fear (that China won't appreciate the risk of making AI Satan or will beat the US to making AI God) and occasional hope (that they might manage to "safety-pill" China and China will find the magic formula to AI alignment). I pulled together some posts previously here: https://awful.systems/post/4103825

I know this is not a novel insight, but all these people in the comments talking about translating HPMOR into as many languages as possible and funding efforts to spread it all over the world... it's bog-standard cult behavior. Substitute "Watchtower magazine" or "Dianetics" for HPMOR to make it even more obvious.

Had someone tell me that they could use the method they use to make these games to make superintelligence, despite the fact that trying this with AI models would be wildly different than doing it with games, and also trying to combine every single AI model into one with this would absolutely obliterate your computer

just milking the hugging face incident for all it's possibly worth

let's continue with the HP theme

One of the most underexplored aspects of Harry Potter's worldbuilding is the color of Lucius Malfoy's hair.

Someone who has never heard of fanfic and just had their mind blown by the worst example of fanfic ever.

I don't like that he's using the language and "philosophy" of racism to hate AI, but that is the most jreg way to approach this.

Jreg also did a video with Aella a couple years back which was an interesting surprise. More recently he was on some Lighthaven podcast. He does a lot of weird things though so idk

q: on the implementation - Is this basically "torturing" the LLM by prompting it with "you are being tortured, lol' in a system prompt?

manifesto: https://jregory.substack.com/p/why-we-are-torturing-clankers

If you called this a schizopost, Jreg would proudly agree with you. Especially the entire "You Are Autistic" section where he invents new and insane flavours of ableism, then proudly declares himself a schizophrenic, for the purpose of hating AI more.

What Would You Do If Your Employer Could Destroy the World

Surely not things you can print in the newspaper. Surely.