Amazon Revises Approach to AI Use by Developers

Amazon has published a memo with updated guidelines on AI usage for engineers and development managers. The document, distributed among employees in September, also redefines their roles in light of advancing technologies. The company urges the use of tools only when necessary and ties performance evaluations to long-term outcomes rather than the volume of generated code.

The company identified token expenditure, AI adoption rates, and the number of code changes or reviews as unreliable metrics: they may increase costs and complicate workflows. This data is suggested to be considered cautiously alongside other metrics. Amazon Web Services Vice President and Distinguished Engineer Mark Brooker stated that Amazon also evaluates "feature release velocity"—the number of launches and successful changes in the production environment.

Amazon separately noted that the lack of benefit from current AI tools for a specific task does not reflect a shortcoming on the employee's part. According to the company, an engineer should be able to explain where the technology is not yet helpful.