OpenAI co-founder and former Tesla AI director Andrej Karpathy announced on May 19 that he has joined Anthropic's pre-training team, where he will build a sub-team using Claude to accelerate core model training research.
Andrej Karpathy shared a practical tip: appending structure your response as HTML to any LLM prompt and viewing the result in a browser. He also outlined a broader vision for the evolution of human-AI interfaces, from text to interactive neural simulations.
Andrej Karpathy shared highlights from his Sequoia Ascent 2026 fireside chat, arguing that LLMs open genuinely new categories of functionality, not just faster versions of what already existed.
Andrej Karpathy says his autoresearch setup reduced nanochat's Time to GPT-2 from 2.02 hours to 1.80 hours. He said the agent explored roughly 700 changes over about two days and found around 20 additive improvements, but the result should still be read as a source claim rather than an independently audited benchmark.
Andrej Karpathy has published autoresearch, a minimal repo that lets AI agents iterate on a stripped-down nanochat training loop overnight. The project turns agent evaluation into a closed-loop research workflow with fixed 5-minute runs, Git branches, and validation-loss-based selection.
Andrej Karpathy highlights the fundamental memory+compute trade-off challenge in LLMs: fast but small on-chip SRAM versus large but slow off-chip DRAM. He calls optimizing this the most intellectually rewarding puzzle in AI infrastructure today, pointing to NVIDIA's $4.6T market cap as proof.
AI researcher Andrej Karpathy argues that programming has fundamentally changed over the last two months, particularly since December when coding agents started actually working. Developers are shifting from writing code to directing and managing AI agents in parallel.
AI researcher Andrej Karpathy argues that LLMs fundamentally change software constraints, excelling at code translation. He predicts large fractions of all software ever written will be rewritten many times over as AI reshapes the programming landscape.
Andrej Karpathy coined a new term for OpenClaw-like AI agent systems: "Claws." Just as LLM agents were a new layer on top of LLMs, Claws provide orchestration, scheduling, persistent context, and tool calls on top of LLM agents.
Andrej Karpathy shared how he vibe-coded a custom health tracking dashboard in 1 hour, then argued that the traditional app store model is becoming obsolete as LLM agents can generate bespoke apps on-demand for individual users.