In their defence, the one thing AI does do well is Code review. But, it makes too many absolutely sh** assumptions when helping customers troubleshooting. I feel the problem is that because its a paid tool, it’s not a good look if they simply acknowledge “we’re guessing, and its a crap guess”