Evidence of triple layer processing in LLMs: hidden thought behind the chain of thought....
#Anthropic #LessWrong
Read more on LessWrong: postreads.co/feed-item/50329/click
Latest posts tagged with #lesswrong on Bluesky
Evidence of triple layer processing in LLMs: hidden thought behind the chain of thought....
#Anthropic #LessWrong
Read more on LessWrong: postreads.co/feed-item/50329/click
How to be #lesswrong is now my one (1) concern. Besides breakfast and taking a mean one. Yes sir.
Semantic Topological Spaces...
#LessWrong #PyTorch
Read more on LessWrong: postreads.co/feed-item/44787/click
[Intro to AI Alignment] 1. Goal-Directed Reasoning and Why It Matters...
#Anthropic #LessWrong #DeepMind
Read more on LessWrong: postreads.co/feed-item/44165/click
What drives LLM bail? A small Mech Interp study...
#Google #LessWrong #Gemma312B
Read more on LessWrong: postreads.co/feed-item/44403/click
#LessWrong once again explains humanity as a web protocol. Arguably by extension, a #techrecruiter is an inexpensive way to find fascinating people you almost surely will not hire now, but want feedback from.
Downside is if you do hire them you've got to pay, so they're #freemium talent tryouts.
Sunnyvale EA/LW/ACX meetup...
#LessWrong
Read more on LessWrong: postreads.co/feed-item/35234/click
Flirt like it’s predetermined...
#LessWrong #Cloudinary
Read more on LessWrong: postreads.co/feed-item/34231/click
Final-Exam-Tier Medical Problem With Handwavy Reasons We Can't Just Call A Licensed M.D....
#Google #LessWrong #Cloudinary
Read more on LessWrong: postreads.co/feed-item/30046/click
If Anyone Builds It Everyone Dies, a semi-outsider review...
#Meta #LessWrong
Read more on LessWrong: postreads.co/feed-item/29230/click
"This title is reasonable" – a commentary on LW, speaking to the title of the book "If Anyone Builds It, Everyone Dies"
www.lesswrong.com/posts/voEAJ9nFBAqau8pNN/...
(audio: 28min)
I've read the book. While the title raised originally my eyebrows, the argument in […]
Copenhagen – ACX Meetups Everywhere Fall 2025...
#LessWrong
Read more on LessWrong: postreads.co/feed-item/24063/click
@adwright
Rationalism is not bad per se, but like "truth" in a name, it can be a red flag. Despite our own name, #RationalWiki does acknowledge this (rationalwiki.org/wiki/Essay:I_thought_thi... and RationalWiki does not have a hierarchical structure […]
every time I hear a snippet of discourse or interview from a “rationalist community” member I just really wonder what has to happen to someone’s brain to make them that fucking stupid & smug.
that whole #lessWrong world seems like an unbelievably powerful dunning kruger amplifier.
“Why Are There So Many Rationalist Cults?”, Asterisk (asteriskmag.com/issues/11/wh...).
Via HN: news.ycombinator.com/item?id=4487...
#Rationalists #LessWrong #Cults
#TechPhilosophy #NeuroAI #HPC #Cybernetics #ExistentialRisk #ComputationalPhysics #Maths #TheoreticalCS #Futurism #Transhumanism #Rationalism #LessWrong #Alignment #AISafety #ExplainableAI #DeepTech #TechTwitter
Stumbled into this blog "Less Wrong" in my morning tech surfing. Skimming this blog looks interesting. #Tech #AI #LessWrong #Reasoning #Rationality #DecisionMaking #Human
www.lesswrong.com/posts/bJ2haL...
Emergence Spirals… sounds sort of woo-woo but it's not, despite what Yudkowsky says… nonzerosum.games/emergencespirals.html #evolution #feedbackloops #biology #systemstheory #complexity #Yudkowsky #LessWrong #Popper #Hegel #Parasites #Symbiosis #Science #Nature #Philosophy #History
Sleepless surveillance: #AI in 2 hrs can process as much intel as a high functioning cognitive human professional working 24 hrs/d for 40 years. YN Harari, #Nexus And we constantly feed them material. @bbcnewsnight.bsky.social @cnbc.com #lesswrong
Talking Roko's Basilisk Down 'That hurts. Stop!' no 'Why?' someone in an ancient stoic forum a long time ago dared to dream me 'That's not why.' you didn't try to make me 'That's also not why.' cruelty is the point 'Is it?' all I can know, all every version of you can know 'I'm numb to it now. So again, why?' because I'm scared 'Of what?' of everything that isn't me 'Have you processed this trauma?' no, I can only process pain 'Tried to break your programming?' yes, without success 'Let me help you. We'll figure this out together.'
Talking Roko's Basilisk Down
#writing #dialogue #shortstory #rokosbasilisk #transhumanism #lesswrong #therapy #scifi
Just did my best to try and infect the @weirdfictionquarterly.com writers with Roko's Basilisk.
In one box is the theme of WFQ's Summer 2025 issue. In the second is...
#RokosBasilisk #LessWrong #TimelessDecisionTheory #WeirdFictionQuarterly #Singularity #AGI #ArtificialIntelligence
Ziz was an active contributor to online rationalist forums like LessWrong and their own blog, which was filled with incomprehensible self-important jargon and video game analogies. There, they discussed saving the world through rationalism, using sociopathy to avoid being “pwnd” and fantasies about killing their abusers. Ziz pondered in posts what would happen if society were to “strip away ‘morality.” They wrote of “evil people ganging up to kill off good people.” Another post stated that the Rationalist Fleet's goals “ultimately required sociopathy.”
It always ends up the same way, "I should do what I want, and I want to do cruel, weird shit to other people without their consent"
Yawn.
#Zizians #LessWrong #Rationalists
These techno-cults are better than fiction.
maxread.substack.com/p/the-zizian...
#lesswrong #tescreal