In fact, automated regression tests done by ai with visual capabilities may have bigger impact than formal verification has. You can have an army of testers now, painfully going through every corner of your software
In practice ends up being a bit like static analysis though, which is you get a ton of false positives.
All said, I’m now running all commits through Codex (which is the only thing it’s any good at), and it’s really pretty good at code reviews.
Will only work somewhat when customers expect features to work in a standard way. When customer spec things to work in non-standard approaches you'll just end up with a bunch of false positives.