I think except for a few isolated use cases, the hype for AI driven testing will probably flare out in the next year or two.
The one use case I’ve seen where machine learning has had undeniable usefulness in testing was a study by mozilla where they had a full test suite that took, like, 100+ hours to run to test firefox builds and they wanted something that decided which tests to run first to get to the first test failure as quickly as possible based upon what code changed. That hit all of machine learning’s strong points, I think.
All in all I’m kind of tired of the ChatGPT hype train though.