Weave's opaque AI coding metrics raise concerns about bias and lack of transparency in employee performance evaluation

Weave analyzes employee AI usage effectiveness, using LLMs to evaluate other LLMs' output—creating a black-box analysis within a black-box system

The tool reduces work performance to a single unexplained metric called 'Code output' per week with no clear methodology or unit definition

Company claims dataset comes from expert-level pull requests but provides no transparency on dataset compilation or whether it covers diverse coding backgrounds

Weave offers no bias assurances or proof despite using these metrics for high-stakes firing decisions based on first-hand accounts

Lack of methodological disclosure and validation raises critical concerns about fairness and reliability of performance evaluations