Like, sure, it came to the same conclusion as the researchers, but they only KNOW that's an accurate explanation because they'd spent years studying it already. The reliability issues we've seen in AI models make it hard to trust without being able to verify.