Andrej Karpathy

Former OpenAI founding member and Tesla AI director, now the field's foremost educator; coined "vibe coding" (vibe-coding). His "Deep Dive into LLMs like ChatGPT" (Feb 2025, 9.7M views, 3.5h) is the canonical general-audience explainer: three training stages (pretraining → SFT → RL) mapped to how children learn from textbooks (exposition → worked examples → practice problems), delivered as a single-take Excalidraw + live-browser screencast with zero hook engineering.

Signature mental models from the lecture: base model as lossy-compression-of-the-internet; assistant as labeler-simulation ("you're not talking to a magical AI, you're talking to an average labeler"); params-vs-context-memory; models-need-tokens-to-think; "RLHF is not RL" (gameable reward models vs indefinite verifiable-domain RL); Swiss-cheese capabilities. Reproduced GPT-2 for $600/24h (lm.c) vs the original $40k.

His closing advice — "use them as tools in the toolbox, check their work, and own the product of your work" — is the same conclusion matt-pocock and dhh reach from the practitioner side: the human owns taste and verification.

Practical sequel, How I use LLMs (Feb 2025, 2.7M views, 2h11m): a tour of ChatGPT, Claude, Gemini, Grok and Perplexity organised by what goes into the context window: tool-selection-by-freshness, thinking models only when the fast answer fails, and verify-like-a-junior-analyst. Personal stack at the time: ChatGPT Pro ($200/mo) as default, Perplexity for search, Claude Artifacts for diagrams, Cursor Composer for code, SuperWhisper on F5 for ~half his queries, and few-shot custom GPTs for learning Korean. He reads books alongside an LLM ("don't read books alone").

Source: report · report