“What if LLMs are mostly crystallized intelligence?” by deep

episode

LessWrong (30+ Karma) not yet transcribed

▲ 0

Transcript

This episode hasn't been transcribed yet Upvotes decide what gets transcribed next — 0 so far. Sign in to upvote

More from LessWrong (30+ Karma)

“Claude Opus 4.5 Achieves 50%-Time Horizon Of Around 4 hrs 49 Mins” by Michaël Trazzi 3 min · 20 Dec 2025

All episodes from LessWrong (30+ Karma) →

Report any issue

Source

Recorded06 May 2026

Publisherfeeds.type3.audio ↗

Handlerec:18285581

Show notes

Summary LLMs are better at developing crystallized intelligence than fluid intelligence. That is: LLM training is good at building crystallized intelligence by learning patterns from training data, and this is sufficient to make them surprisingly skillful at lots of tasks. But for a given capability level in the areas they’ve trained on, LLMs have very weak fluid intelligence compared to humans. For example, two years ago I thought human-level SAT performance would mean AGI, but turns out LLMs can do great at the SAT while being mediocre at lots of other tasks. I’m not saying LLMs are just parrots (that's dumb).[1] There's a continuity between crystallized and fluid intelligence. At the extreme “crystal” end we have shallow locally-valid heuristics. Pure pattern matching. Now-largely-debunked “stochastic parrot” hypothesis.At the extreme end of “fluid” you have a cross between an idealized consultant, a Renaissance man, and MacGuyver. A deep world model and general reasoning, able to come to grips with any particular environment and problem, and to invent new tools and concepts on the fly.Some other ways to gesture at this: what n-gram of Markov chain you’d need to capture a behavioral pattern; number of tasks the pattern is [...] ---Outline:(00:10) Summary(02:47) Implications for AI futures(04:54) We should check if this is true!(05:44) Modeling worlds where AI progress is hungry for domain data(06:15) What types of areas see progress in this model?(09:38) There are also stories for how advanced AIs could route around data bottlenecks:(10:36) Which concrete domains see progress?(12:03) Implications for AI takeoff(12:08) While it lasts, weak fluid intelligence is great news for alignment risk(12:58) A key bifurcation point: can AIs revolutionize AI R&D, or merely speed it up?(14:51) Is this the world we live in?(16:09) How can we test this hypothesis? The original text contained 1 footnote which was omitted from this narration. --- First published: May 5th, 2026 Source: https://www.lesswrong.com/posts/Zxw3ZcmSdndpQyJ6M/what-if-llms-are-mostly-crystallized-intelligence --- Narrated by TYPE III AUDIO.

Ask this episode

“What did the council decide on item 42?” — answered with the speaker and the moment.