AI 2027 forecasters pull AGI estimates forward
The AI 2027 forecasting team, led by Daniel Kokotajlo and Eli Lifland, has shifted their median AGI expectations 1.5 years earlier, now targeting 2027 or 2028. This update reverses a late-2025 "slowdown" trend, citing the rapid maturation of autonomous coding agents and faster task-completion time horizons in frontier models like Gemini 3 and GPT-5.2.
The sudden reversal of the "data wall" narrative suggests that algorithmic efficiency is outpacing hardware constraints. Coding agents like Claude Code are becoming leading indicators for AGI, proving that AI-assisted AI research is already providing tangible speedups. The shift is supported by the physical feasibility of "human-sized" compute clusters using NVIDIA Blackwell and Vera Rubin hardware, while Daniel Kokotajlo's decision to cut the "Automated Coder" milestone requirement highlights a massive jump in agentic reliability.
DISCOVERED
9d ago
2026-04-03
PUBLISHED
9d ago
2026-04-02
RELEVANCE
AUTHOR
socoolandawesome