Study: AI coding agents generate more code, not more software

A Harvard study using Jellyfish analytics spanning over 700,000 employees and 300 million work events finds AI coding efficiency gains are absorbed by code-review bottlenecks. Firms saw little evidence of higher software output or lower employment, with longer reviews and more revised pull requests.

• Anthropic launches tools to show AI spending value2026-10-28
• Harvard study: men more likely to fake AI knowledge2026-10-14