I think "gambling" is a bit too strong, but there is a real shift in how we evaluate correctness. With traditional coding, you reason step by step and with AI-assisted code, you're often validating outputs after the fact.
The risk isn't randomness per se it's over trusting something that looks correct. The skill ceiling is moving from "can you write it" to "can you reliably verify it"
The risk isn't randomness per se it's over trusting something that looks correct. The skill ceiling is moving from "can you write it" to "can you reliably verify it"