Amazon warns engineers against AI ‘tokenmaxxing’

Amazon Amazon.com, Inc. ↘ 4.25% has told its software engineers that heavy AI use is not the same as productive work. The company said tokenmaxxing, the practice of using AI tools as much as possible with little regard for the results, is out of step with how it wants engineers to work. The guidance came in a memo CNBC reported was circulated internally in September and posted to Amazon's website on Tuesday, October 27. It revises how the company defines the roles of software development engineers and managers.cnbc.com

Terk Terkowitz, vice president and distinguished engineer at Worldwide Amazon Stores, wrote the memo. It says Amazon has entered "the era of the Renaissance Developer," in which engineers blend AI tools with human judgment and domain expertise, according to Fortune.fortune.com

Metrics called "deeply flawed"

The memo says token usage, AI tool adoption rates and counts of code changes or reviews are "deeply flawed" measures because "they are disconnected from what matters" The Register reported. It warns that treating them as targets could raise costs, since "tokenmaxxing is at odds with frugality," and that measuring lines of code works against Amazon's "invent and simplify" principle, according to the full text published by the Times of India.theregister.com+1

Amazon is not dropping those metrics entirely. The memo says they "correlate well with the outputs that matter." It asks managers to look at the bottom 10% of teams by AI adoption to understand why they lag, and to check whether the top 10% are delivering more to match, The Next Web reported.thenextweb.com

"A finding, not a failing"

The memo says AI will stay central. "We are certain that whatever future best practices are, AI will be at the center of software engineering practice," the company wrote, according to CNBC. It also tells engineers not to force the tools where they don't fit. "If your work does not benefit from current AI tooling, that is a finding, not a failing," it says, adding that an engineer who can explain where AI does not yet help "is demonstrating the judgement we are asking for" Fortune.fortune.com+1

No token caps

An Amazon spokesperson told CNBC that the company has not capped employees' token usage. Amazon shut down an internal leaderboard in May after it found employees were tokenmaxxing to climb the rankings. Fortune reported the leaderboard was called Kirorank and that some workers had run AI agents on meaningless tasks to raise their token counts Fortune.cnbc.com+1

Marc Brooker, a vice president and distinguished engineer at Amazon Web Services, said the company tracks "feature velocity" to judge whether AI is making engineers more productive, Quartz reported. Microsoft sent a similar message to its own developers in August, when executive Jay Parikh said "tokenmaxxing is not what we are optimizing for" The Register.qz.com+1

"What we don't want to see is wasted effort where we are going and spending those tokens solving problems that don't matter," Brooker said Quartz.qz.com