Developers refuse to work without AI coding tools, but research suggests the productivity gains may not be real and could increase maintenance costs

In February 2026, METR found that most developers won't work on even limited tasks without AI anymore. The lab had planned to repeat earlier research measuring AI coding productivity, but developers refused to participate in the study without AI access.

Earlier research from 2025 showed that while developers perceived AI made them faster, it actually slowed them down overall—AI generated code quickly, but developers spent extra time finding and fixing errors, steering the AI, and waiting for it to complete tasks.

Amazon shut down its internal token-tracking leaderboard Kirorank after employees gamed it by using AI agents excessively and running up costs. Uber spent its entire 2026 AI budget within the first four months, with COO Andrew Macdonald saying the spending hadn't led to measurable increases in projects or productivity.

CodeRabbit found that AI produced 1.7x more problems than human code in open source pull requests. A Singapore Management University report published in April warned that AI-generated code can introduce long-term maintenance costs into real software projects.