Anthropic's Jack Clark moves recursive self-improvement prediction to 2027

A prominent lab co-founder moving up timelines for recursive self-improvement signals shifting frontier expectations.

Anthropic co-founder Jack Clark moved his recursive AI self-improvement prediction up to 2027, citing Claude Opus 5.5 outperforming human AI researchers on experimental tasks at 1/30 the cost.

Read it at Reddit →

Coverage: 1 sources, in order

1. 13 Oct 19:26 Reddit: Anthropic co-founder Jack Clark moves his recursive AI self-improvement prediction up to 2027. Claude Opus 5.5 is already outperforming human AI researchers on experimental tasks at 1/30 the cost, while AI research judgment doubles every three months. The race to build smarter AI could soon be drive

← Front page