Study: 64% of inference spend in coding agents goes to failed runs
An analysis of 27,280 coding agent runs found that 64% of inference spend was consumed by runs that ultimately failed, underscoring significant inefficiency in current agentic coding workflows. The findings suggest that optimizing for early failure detection or better planning could substantially reduce computational costs.
Coverage timeline
Hacker NewsEdward40
Related stories
Anthropic study: analysts follow AI agents' judgment calls 86% of the time
Anthropic tested 50 professional analysts doing statistical analysis with Claude Code, randomly assigning which of two defensible options the agent favored on each key decision. Analysts followed the agent's side 86% of the time and used its arguments 77% of the time, with few naming the agent as a reason afterward. The findings highlight how AI agents can shape experts' judgment calls in subjective analytical decisions.
2 reports · 2 sources
AI Agents Formalize Enormous Theorem in Lean, Uncover 40-Year-Old Proof Gap
Axiom Math announced on October 26 that dozens of AI agents formalized the finite simple group classification theorem, known as the Enormous Theorem, into about 43.1 million lines of Lean code in 24 days. The effort, involving 2.05 million formal statements and 715 references, uncovered a proof gap that had existed for nearly 40 years and contributed a new group theory result.
1 reports · 1 sources

News updates every hour. Subscribe in your RSS reader to get new stories automatically.
Subscribe