According to this study, use of generative AI for coding increases task completion time by 19%. One of the factor likely contributing to slowdowns is low AI reliability:

- Developers accept < 44% of AI generations

- Majority report making major changes to clean up AI code

- 9% of time spent reviewing/cleaning AI outputs

Also, developers report AI doesn't utilize important tacit knowledge or context about the repository.

Does this resonate with your experience? Let me know in the comments.