Amazon publicly posted guidance for software engineers on Oct. 27 that favors customer and business results over maximizing AI use, CNBC reports. The guidance had circulated internally in September. The Register identifies Amazon VP and distinguished engineer Terk Terkowitz as the note’s author.
The guidance calls token usage, AI-tool adoption rates, and counts of code changes or reviews “deeply flawed” ways to measure teams’ AI use, according to CNBC. It does not tell engineers to discard those numbers: Amazon says to consider them “with care and judgment” alongside other data. Its stated preference is to “measure, and be measured by, long-term outcomes.”
Treating such metrics as a target rather than a measure “could” increase costs and complexity, The Register quotes the note as warning. CNBC reports that Amazon shut down an internal leaderboard in May after finding employees were consuming AI tokens to climb its ranks.
Amazon Web Services VP and distinguished engineer Marc Brooker told CNBC that Amazon has considered “feature velocity,” including launches and successful changes in production, when assessing how AI improves engineers’ productivity. He framed spending on tools, tokens, time and human effort around important customer and business problems, while allowing room to experiment.
The guidance is not a call to stop using AI. Amazon says AI will be central to software engineering, and a spokesperson told CNBC the company has not imposed caps on employees’ token usage. It also asks engineers to recognize where the tools fall short: “If your work does not benefit from current AI tooling, that is a finding, not a failing,” CNBC quotes the note as saying.
Comments