Reaction
The case for nonstop AI-agent testing at frontier labs
A user argues frontier labs should run thousands of agents constantly testing every feature across all combinations.
TLDR
A user calls buggy software from frontier labs inexcusable in the era of “super human models.” They argue that thousands of agents should constantly try every feature in every combination, and that users should never see even a rare bug.
Combined views
—
1 Source, first seen ago
