Well I guess if the candidate would be a little be stronger and actually trying to reason with the LLM about the decision it suggested, he would be better prepared and maybe got away with his claims.
Or as current best chess player Magnus Carlson said, "if I would cheat, you would never know". Meaning very strong candidates will get away with flexing the truth with AI. But this means maybe, you shouldn't look for a perfect fit. Or check his merit by spending time and money to get in touch with his old companies.
Yeah.. but if he didn't actually work exactly on it, but took the effort to learn from coworkers (or LLMs or google or wherever) and is able to answer my questions on what he did, and more importantly on why he decided to do something a certain way and not some other way, then he/she must have spent considerable amount of time actually learning about it and figuring things out. So I'd still hire him/her. The trouble is most people who embellish are either not competent to go deep enough to learn, or think that they can get away with some superficial knowledge of it.
Or as current best chess player Magnus Carlson said, "if I would cheat, you would never know". Meaning very strong candidates will get away with flexing the truth with AI. But this means maybe, you shouldn't look for a perfect fit. Or check his merit by spending time and money to get in touch with his old companies.