Top forecaster Eli Lifland moves expert-level AI timeline from 2034 to 2028, citing METR report
birchlse · x · 2026-11-11
Eli Lifland, a careful professional forecaster at AIFP, has moved his estimate for top-expert-dominating AI from 2034 to 2028, with automated AI research going into overdrive in 2027.
- Daniel Kokotajlo says the new METR report implies an intelligence explosion in 2027, five months ahead of the AI 2027 scenario, adding he is "scared, sad, and angry"
- The forecast table shows authors' median estimates for automated AI research clustering between May–Nov 2027, and TED-AI (AI better than top human experts at everything) between Sep 2027–Sep 2028, with 35–65% odds by end of 2027
- These forecasters previously made seemingly preposterous calls on frontier lab capex, revenue, and capability growth that proved correct; on average they put civilization-scale catastrophe or permanent disempowerment odds at a coin flip or worse
More from AGI Musings
- OpenAI and Upwork study freelance job postings as early signals of AI changing work — Exp_Mark · 2026-11-11
- As AI grows more powerful, researchers reflect on agency, happiness and meaning — prajdabre · 2026-11-11
- Antitrust risk as a blocker to AI safety coordination: real concern or convenient excuse? — lxrjl · 2026-11-11
- Blaise Aguera y Arcas on symbiogenesis: automation accelerates trading self-sufficiency for collective capability — blaiseaguera · 2026-11-11
- Blaise Aguera y Arcas: master-servant AI framings are unhealthy for both parties — blaiseaguera · 2026-11-11
- Swarm riders: a new class of agent commanders spending 1000x more tokens than median engineers — JensHonack · 2026-11-11