This is actually a very good point. Although it's indeed not hard to imagine AI being far better at estimating the complexity of a potential solution and warnin

This is actually a very good point. Although it's indeed not hard to imagine AI being far better at estimating the complexity of a potential solution and warning the user about it. For example in chess AI is already far better than humans. Including on tasks like evaluating positions. Admittedly, I use "AI" in a broad sense here, despite the article being mostly focused on LLMs.