METR trial finds early-2025 AI tools slowed experienced developers
A randomized study found experienced open-source developers took 19 percent longer with the tested AI tools on familiar repositories.
0 comments · 0 votes
Sign in to join the discussion →
No comments yet. Start the discussion.
Why it moved the index
The controlled result constrained claims that contemporary systems had already surpassed experienced humans in complex, context-heavy software work.
Assessment history
-
R1
Away 38 · confidence 92
Initial source-backed launch assessment
11 Aug 2026
Share this page
-
DoomBench assesses “METR trial finds early-2025 AI tools slowed experienced developers” as evidence moving away from doom, with magnitude 38 and confidence 92 out of 100 in the human resilience category.
-
The DoomBench assessment of “METR trial finds early-2025 AI tools slowed experienced developers” is based on reporting from METR and records the editorial rationale, source quality, attribution, and revision history.
-
DoomBench summarizes “METR trial finds early-2025 AI tools slowed experienced developers” as follows: A randomized study found experienced open-source developers took 19 percent longer with the tested AI tools on familiar repositories.
https://www.doombench.com/news/metr-finds-ai-slowed-experienced-developers