Maybe someone who actually works at FAANG can weigh in, but I would think that one of your best bets would be getting into one as a general SWE and then transitioning to AI/ML internally after a year. I recall Google even having some sort of internal program that encouraged this. Getting into Google is a moonshot, but it's possible to do so with no prior professional programming experience if you put in a ton of effort AND get lucky. Amazon seems willing to train too, based on the following experience I had:

I'm an iOS engineer without a STEM background, and I've been contacted by Amazon recruiters for entry-level ML/AI positions. I thought it was weird, but they said they've hired a few people with iOS backgrounds and no prior ML/AI experience who are now excellent ML engineers. I backed out because I knew I would fail the interview process at this point, but it's something for me to think about for the future.