I don't exactly see it this way - "You can only do this by doing actual discovery interviews with prospects in an unbiased way where you explore their experience around specific problem areas without ever mentioning your brilliant solution - as proposing a specific solution is likely to bias your subject."
Sometimes the entrepreneur needs to expand the possibility space, rather than just addressing an "unmet need". There's the Jobs-ism about figuring out what the customer is going to want before they do. Thiel also talks in Zero to One about some of the biggest innovations not spawning from customer feedback or lean methodology.
I wouldn't say that GPT addressed a problem I had per se. It just created a whole new set of activities I wanted to try.