> The second portion of your statement is either confusing (something unsaid) or untrue (you are still ultimately on the hook).
You're missing the point.
tptacek is saying he isn't the one who needs to fix the issue because he can just reject the PR and either have the AI agent refine it or start over. Or ultimately resort to writing the code himself.
He doesn't need to make the AI written code work, and so he doesn't need to spend a lot of time reading the AI written code - he can skim it for any sign it looks even faintly off and just kill it if that's the case instead of spending more time on it.
> My previous point stands - if it was that cut and dry, then a (free) script/library could generate the same code.
There's a vast chasm between simple enough that a non-AI code generator can generate it using templates and simple enough that a fast read-through is enough to show that it's okay to run.
As an example, the other day I had my own agent generate a 1kloc API client for an API. The worst case scenario other than failing to work would be that it would do something really stupid, like deleting all my files. Since it passes its tests, skimming it was enough for me to have confidence that nowhere does it do any file manipulation other than reading the files passed in. For that use, that's sufficient since it otherwise passes the tests and I'll be the only user for some time during development of the server it's a client for.
But no template based generator could write that code, even though it's fairly trivial - it involved reading the backend API implementation and rote-implementation of a client that matched the server.