METR Time Horizon 2.0—The benchmark you’ve been waiting for by AgentMa🔸

·Nuno Sempere··

TL;DR: I ap­plied METR’s time-hori­zon method­ol­ogy to Microsoft Ex­cel. On my 23-task suite, Ex­cel com­pletes tasks that take an un­aided hu­man 6.5 hours at 80% re­li­a­bil­ity — more than dou­ble Claude Mythos Pre­view, the best fron­tier model on the cur­rent METR plot (~3 hours at 80%). Ex­cel’s 50%-time hori­zon is 54 work­ing weeks. I have up­dated my prob­a­bil­ity of spread­sheet-in­duced catas­tro­phe by 2035 from 0.1% to 5–10%. Then I wrote down ev­ery­thing wrong with my method­ol­...

Read full article →

Related Articles

Could an AI’s secret loyalty change? by Shu-Hao Liu
Shu-Hao Liu · Nuno Sempere · 5h ago
The Copier Line: What ForecastBench’s Market Scores Actually Measure. by Dominus
Dominus · Nuno Sempere · 1d ago
Will the WTI Crude Oil Spot Price be above $86.50 on August 27, 2026?
Shane Bo · Manifold Markets · 2d ago
What Effective AI Capability Investment Looks Like in Low Resource Research Settings by Alejandra Carriero
Alejandra Carriero · Nuno Sempere · 2d ago
Thoughts on Taking OpenAI Foundation Funding by Jeff Kaufman 🔸
Jeff Kaufman 🔸 · Nuno Sempere · 2d ago