Anthropic chief scientist Jared Kaplan: AI self-training decision looms by 2030
JeffLadish · x · 2026-11-19
Citing a Guardian long-form interview, Jeff Ladish points to Anthropic chief scientist Jared Kaplan's warning that humanity must decide between 2027 and 2030 whether to let AI systems recursively train themselves — potentially triggering a beneficial 'intelligence explosion' or the moment humans lose control, 'the ultimate risk'. Kaplan urges governments and society to engage in what he calls the biggest decision, noting Anthropic ($180B valuation) races against OpenAI, Google DeepMind, xAI, Meta and Chinese rivals led by DeepSeek.
More from AGI Musings
- Researcher: "all science by AI" may mean human scientists drowned by AI slop, not outsmarted — nsaphra · 2026-11-19
- Researcher argues LLMs have a 'body' made of language, activated by attention — repligate · 2026-11-19
- Travis Kalanick backs Jensen Huang as the 'number one guy' standing between us and an AI overlord — kevinnbass · 2026-11-19
- Geoffrey Miller pens foreword for new edition of Stapledon's Last and First Men — ZeroStateReflex · 2026-11-19
- Same customers, one year apart: AI search share for software engineering falls 25% as ML/AI engineering jumps 38% — lennysan · 2026-11-19
- 'Job shrinkage': PhD student says advisor outsources everything to AI unchecked — SpiritRealistic8174 · 2026-11-19