"But the gen-a.i. assistants are guided by tests!"

So here's a thing about tests: when a program passes a set of tests, that's useful because it rules out some ways in which the program's behavior could have diverted from expectations that followed from your understanding of the program.

But you do need to have a clear understanding of the program. And as your "productivity" increases (as you produce more & more code in less & less time), your understanding fades.