Image created with gemini-2.5-flash-image with claude-sonnet-4-5. Image prompt: A polished wooden chess board in mid-game with a white knight piece suspended in air above it, multiple faint glowing squares below showing possible moves, warm side lighting creating long shadows, intimate close-up perspective emphasizing the moment of difficult choice, photorealistic style with shallow depth of field.
Andrej Karpathy — “We’re summoning ghosts, not building animals” – YouTube https://www.youtube.com/watch?v=lXUZvyajciY
On Dwarkesh Patel’s Podcast With Andrej Karpathy https://thezvi.substack.com/p/on-dwarkesh-patels-podcast-with-andrej
.@karpathy says that LLMs currently lack the cultural accumulation and self-play that propelled humans out of the savannah: Culture: > “Why can’t an LLM write a book for the other LLMs? Why can’t other LLMs read this LLM’s book and be inspired by it, or shocked by it?” Self https://x.com/dwarkesh_sp/status/1980333945385562176
The most interesting part for me is where @karpathy describes why LLMs aren’t able to learn like humans. As you would expect, he comes up with a wonderfully evocative phrase to describe RL: “sucking supervision bits through a straw.” A single end reward gets broadcast across https://x.com/dwarkesh_sp/status/1979259041013731752
There are at least a dozen models of developers interacting with AI to code and since we don’t have a taxonomy of them, everything is “”vibecoding”” which can mean autonomous coding or 10x improvements in performance or the production of complete slop, depending on who says it.”” / X https://x.com/emollick/status/1980078073589661956
Apple employees have ‘concerns’ over Siri performance in early iOS 26.4 builds: report – 9to5Mac https://9to5mac.com/2025/10/19/apple-employees-concerned-by-early-ios-26-4-apple-intelligence-sir-version/
An Opinionated Guide to Using AI Right Now https://www.oneusefulthing.org/p/an-opinionated-guide-to-using-ai
nanochat now has a primordial identity and can talk a bit about itself and its capabilities (e.g. it knows it’s nanochat d32 that cost $800, that it was built by me, that it can’t speak languages other than English too well and why, etc.). This kind of customization is all done”” / X https://x.com/karpathy/status/1980665134415802554
One cannot show that turbojets are safe before actually building turbojets and carefully refining them for reliability. The same goes for AI.”” / X https://x.com/ylecun/status/1981360519442321451
A lot I like & some I don’t in this paper: Like: Clear definition of AGI, diverse authors, shows jaggedness, tracking metrics over time (huge leap from GPT-4 to GPT-5) Dislike: AGI defined as replicating a model of human cognition, benchmarks are scattershot, narrow view of AI https://x.com/emollick/status/1978874737892667718
A focus on AGI (whatever it is) obscures the fact that we have increasing evidence from early results like GDPval that today’s AI models are good enough to create major transformations over 5-10 years as companies figure out how to deploy them and integrate them into processes. https://x.com/emollick/status/1979287090962788531
Given a very intelligent AI that can solve very hard problems, what profound question will you ask? https://x.com/emollick/status/1978995377488486803
Postponing Grokipedia v0.1 launch to end of week. We need to do more work to purge out the propaganda.”” / X https://x.com/elonmusk/status/1980340232194257090
Our Approach to Teen AI Safety: Empowering Parents, Protecting Teens https://about.fb.com/news/2025/10/teen-ai-safety-approach/
Human-centered AI | Microsoft Copilot Blog https://www.microsoft.com/en-us/microsoft-copilot/blog/2025/10/23/human-centered-ai/
Tomorrow https://x.com/dwarkesh_sp/status/1978885626255839419
Until recently, datacenters were able to simply repurpose the carcass left over from America’s deindustrialization. Honestly a compelling ode to capitalism. https://x.com/dwarkesh_sp/status/1981113646836306030
