Amazon tells engineers to stop ‘tokenmaxxing’ to look productive
Amazon has published new engineering guidelines warning developers and managers against ‘tokenmaxxing,’ arguing that AI usage metrics are ‘deeply flawed’ unless tied to long-term outcomes. The move follows similar spending concerns at Uber and Accenture.
Amazon has issued new guidelines to its engineers, urging them to stop prioritizing token counts when evaluating AI usage. The company warns that simply consuming more tokens does not equate to delivering better results. In a memo posted on October 27, Amazon clarified that metrics such as AI adoption rates, token consumption, code changes, and code reviews should not be used as primary performance indicators if they lack a connection to actual outcomes.
Instead, the company advocates for measuring long-term results to assess the true value of AI implementation. Amazon remains committed to utilizing AI within its operations, stating that the technology will reshape the pace and scale of everyday work. However, the company emphasizes that engineers should avoid 'tokenmaxxing'—the practice of intentionally inflating AI usage to appear more productive.
Amazon suggests evaluating both the teams that use AI the least and those that use it the most to determine if adoption is translating into improved delivery. The guidelines also caution against using AI tools unnecessarily, as such behavior may indicate a lack of judgment rather than a reluctance to adopt AI technology.
Written by urgent.news from Tom's Hardware's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
