Amazon warns engineers against measuring success by AI use

Amazon is urging its engineers to judge work by its results, rather than by how much they use artificial intelligence (AI). In a memo to employees, the company called measures such as AI token use and code generation “deeply flawed” ways to gauge a team’s success.

The memo also cautioned against “tokenmaxxing,” or chasing high token use. It said the company’s most prolific engineers show that neither the developer role nor a worker’s output is defined by AI use.

Written by Amazon vice president and distinguished engineer Terk Terkowitz, the memo is titled “An update from Amazon engineering leadership on the role of the developer in the age of AI.” It describes a “Renaissance Developer” who combines AI tools with human judgment and domain expertise to build new things.

Terkowitz wrote that AI is changing software engineering rapidly and fundamentally. The memo says Amazon wants teams to focus on effects on customers, businesses and colleagues as it evolves software development engineer roles. It also asks managers to ensure teams understand AI’s benefits and encourages engineers to stay closely connected to customers to understand their needs.

Terkowitz said the effects of AI on software engineering are only beginning to emerge, and that AI will be central to future software engineering practices.