OpenPrediction

By end-2027, models will frequently outperform experts on many tasks

“By the end of 2027, models will frequently outperform experts on many tasks.”

Julian SchrittwieserAI researcher at Anthropic; formerly Google DeepMind (AlphaGo, MuZero)Said · Source: julian.ac

Context: "Failing to understand the exponential, again" (Sep 27, 2025).

How we judge it

Our interpretation: by 31 Dec 2027, a model clearly beats (not just ties) industry experts on a majority of tasks in GDPval or an equivalent evaluation.

Resolution criteria are our interpretation of the claim, written before grading. Methodology

Said
Resolve by
Due in 449 days
Verdict
Open
Stated probability
None stated
Specificity
1 of 3 (vague or hedged)

Share this scorecard

Post on X LinkedIn Share image (PNG)