Yudkowsky Contra Christiano On AI Takeoff Speeds

A clear, careful explainer/adjudication of the Yudkowsky-Christiano fast-vs-slow-AI-takeoff debate (part of the Yudkowsky-Contra series). Lays out Paul's gradualist case (4-year-doubling-before-1-year; smooth historical curves) vs Eliezer's discontinuity case (secret-sauce, chimp->human, deceptive-AI, recursive self-improvement), the chimp-human nitpick, 'reference-class tennis', the sigmoid/Gwern point, and reaches a balanced verdict (Paul absolutely, Eliezer directionally; prepare for either).

Quality 77 · Excellent. Strong: one of the better popular explainers of a genuinely hard AI-safety debate, with Scott's own analysis woven through; a signature Contra-series piece.

Claude’s paradigm shift 52 · Moderate. Notable: crisp synthesis of both positions + the sigmoid/why-it-matters framing, building on the participants' arguments.

Real-world impact 3 · Moderate. Influential AI-takeoff-debate explainer widely read in AI-safety circles. =3.