a) don't overestimate LLM
b) cache the right answer so you don't call the LLM api everytime
c) fast enough
idk if we can say "fast", maybe more of a time saver. the total spent analysing UI and writing tests will be superior of writing "playsmart" tests + runtime.
agreed. one may also say cheaper than having one full time employee writing tests and updating them.