Man, @karpathy liked this post, but I'm scared now
AGI
-
LLMs World Models Gap: Ongoing Critique and Arguments
By
–
among other places on facebook immediately after my deep learning is hitting a wall paper where i laid out the arguments in question, and also for years here on twitter, eg in 2019 on X calling my critique that LLMs lacked world models a “rear-guard action” see my recent
-
Scaling Limits Debate: LLMs Won’t Achieve AGI Claims
By
–
Wow. Just wow. @ylecun taking credit for my March 2022 argument that scaling would hit a wall and that LLM would not bring us to AGI–after he initially attacked me for saying it and continued to promote them right up until ChatGPT ate has lunch–has to among the most
-
Opus outperforms GPT-5-Pro on ARC AGI reasoning tasks
By
–
On ARC AGI 2, Opus is just a smarter model and I think we see such big climb not because it is so good at thinking, but rather without thinking it is nearly impossible for it to do these puzzles well (same for me). E.g. GPT-5-Pro is below Opus 4.5 thinking 16k, but there's no way
-
Scaling Alone Cannot Reach AGI: Vindication of Longstanding Critique
By
–
It is one thing for everyone from @sama to @ylecun to @elonmusk and the Twitterverse to have attacked me literally for years for saying that scaling alone would not get us to AGI.
— Gary Marcus (@GaryMarcus) 26 novembre 2025
Another to pretend now that that never happened — now that I have largely been proven correct. pic.twitter.com/n77ChV5lrQIt is one thing for everyone from @sama to @ylecun to @elonmusk and the Twitterverse to have attacked me literally for years for saying that scaling alone would not get us to AGI. Another to pretend now that that never happened — now that I have largely been proven correct.
-
AI Agents Achieve Direct Mind-to-Mind Communication Framework
By
–
AI agents could skip language entirely and communicate mind to mind. This work introduces thought communication, a latent variable framework that identifies shared and private thoughts across agents and recovers the global structure of who shares what. The approach extracts
-
AI breakthroughs for YouTube channel growth
By
–
9. "What is the minimum viable breakthrough?" Not minimum viable product. Minimum viable breakthrough. "I want to grow a YouTube channel. What is the minimum viable breakthrough?" AI finds the foundational leap that unlocks growth.
-
AI generates optimal onboarding solution first
By
–
4. "What would the optimal solution look like if cost didn’t exist?" Constraints ruin creativity too early. "I need to redesign onboarding. What would the optimal solution look like if cost didn’t exist?" AI generates the ideal version first, then works backward.
-
AI prompts inspired by Elon Musk’s thinking framework
By
–
I used Elon Musk's actual thinking framework as AI prompts. It's the closest thing to having a billionaire engineer rip apart your ideas and rebuild them from physics. Here are the 15 prompts that changed how I solve problems:
-
VLMs Learn Visual Reasoning with COVT Paper
By
–
I just read this paper called "Chain-of-Visual-Thought (COVT)" and it basically teaches VLMs to see and think at the same time not in text, but in continuous visual tokens. Here’s the wild part: Instead of forcing models to reason through words (which destroys all the