back
Get SIGNAL/NOISE in your inbox daily
Building on the recent empirical work of Kwa et al. (2025), I show that within their suite of research-engineering tasks the performance of AI agents on longer-duration tasks can be explained by an extremely simple mathematical model — a constant rate of failing during each minute a human would take
Recent Stories
Jan 19, 2026
App Store apps are exposing data from millions of users
An effort led by security research lab CovertLabs is actively uncovering troves of (mostly) AI-related apps that leak and expose user data.
Jan 19, 2026Stop ignoring AI risks in finance, MPs tell BoE and FCA
Treasury committee urges regulators and Treasury to take more ‘proactive’ approach
Jan 19, 2026OpenAI CFO Friar: 2026 is year for ‘practical adoption’ of AI
OpenAI CFO Sarah Friar said the company is focused on "practical adoption" in 2026, especially in health, science, and enterprise.