Research and real-world data suggest AI coding tools may be hurting developer productivity despite widespread adoption. METR's 2025 study found AI actually slowed developers down, yet developers now refuse to work without it. Amazon shut down its internal AI token leaderboard (Kirorank) after employees gamed it, and Uber burned through its entire 2026 AI budget in four months with no measurable productivity gains. The phenomenon of 'tokenmaxxing' — using token consumption as a productivity proxy — is producing wrong incentives. Studies show AI-generated code introduces 1.7x more bugs than human code, with 44% of tokens spent fixing AI-generated bugs. Researchers recommend treating AI output like junior developer code: review everything and keep humans responsible for architecture and security.

4m read timeFrom thenextweb.com
Post cover image

Sort: