METR Survey: Developers Report AI Doubles Their Value to Organizations
Technical employees report AI makes them twice as valuable, yet studies show mixed productivity results and rising maintenance costs. Companies report heavy spending on AI-generated code fixes.
theweek.comA May 2026 METR survey found that technical employees perceived AI made them twice as valuable to their organizations. The self-reported data came after METR researchers could not repeat a 2025 experiment because developers refused to work without AI tools. In February 2026 METR had published findings that most developers would not work even on a limited number of tasks without AI.
When researchers attempted to update the 2025 study measuring task completion times with and without AI assistance, participants declined because they did not wish to work without AI. The 2025 METR study had measured how much time open source developers took to complete tasks by hand versus with AI.
Developers in that study reported that AI made them more productive, but the actual measurements showed AI slowed them down as they spent extra time finding and fixing errors.
Tokenmaxxing, using the number of tokens a person uses as a proxy for productivity with AI, became the trend of 2026. Amazon shut down its internal token-tracking leaderboard called Kirorank after employees gamed it by using AI agents excessively. Uber exhausted its entire 2026 AI budget within the first four months of the year.
Uber COO Andrew Macdonald said on a podcast that excessive AI spending had not led to a measurable increase in projects or productivity. Companies are spending 44% of their tokens on bug fixes for AI-generated code, according to Aiswarya Sankar. 7 times more problems than human code.
Researchers from Singapore Management University published a report in April 2026 warning that AI-generated code can introduce long-term maintenance costs into real software projects. The researchers recommend that programmers know what tasks AI does and does not do well, implement strong quality assurance systems designed for AI, carefully review AI's work, and have humans handle software architecture and security design.
Cognition founder and CEO Scott Wu suggested that AI coding agents can handle fixing code generated by AI.
Wu rates the skill of Cognition's Devin AI coding agent between junior and mid-level programmer depending on the task. Programmer and author James Shore wrote in a blog post that went viral on Hacker News: "You write code twice as quick now? Better hope you've halved your maintenance costs.
Otherwise, you're screwed.
