Questions about Playwright/Automation testing

1) Are AI based test automations tools even useful? I see a number of them coming up but I still don't find them to be very good yet. 2) What are the hardest problems faced while writing playwright/automation tests? 3) Are tests written using playwright flaky? How are they solved today?