Hacker News Reader

I think it'll be much more likely that the AI writes the tests (the boring stuff) for the buggy code I write.

I can see the AI suggesting and fixing syntax in the tests. Determining their semantics, not without true AGI.