Comment by esafak

Comment by esafak 2 days ago

0 replies

There is an attribution problem because we don't see the interaction the engineer had with the AI before submitting the work. Maybe the prompt was good, and there was a back-and-forth as the user found mistakes and asked for corrections. Maybe not. We are left to guess. A user who does not steer the AI adds no value at best, and likely makes things worse by creating work for others. There is a product opportunity here for coding agents to check in this work with something like `git notes`. This helps users claim the value they add.