Mark Cuban and Marc Andreessen got into it on X last week.

The fight: is AI now better than doctors?

New data showed AI got it right more often than human doctors in blind tests, about 20-25% of the time versus 10%. Andreessen said it's a huge win for AI. Cuban said: cool story, but who cares if the insurer still finds a way to deny your claim?

Here's each side's argument, to the best our pea brains can explain it:

Andreessen's side: AI is already a better doctor than most humans. The data backs it up, so why not use it? AI doesn't get tired, doesn't forget rare conditions, and can pull from way more medical info than one doctor ever could. If it's more accurate, keeping it out of medicine could actually cost lives, especially in places with doctor shortages.

Cuban's side: Cool diagnosis, but so what? A great diagnosis means nothing if the insurer still finds a way to deny the claim. And this isn't hypothetical. NAIC found 84% of health insurers already use AI, and 12% use it for prior-auth denials. Cigna, UnitedHealthcare, and Humana are all being sued over it right now. Meanwhile, fewer than 1% of denied ACA claims ever get appealed. Most people just take the "no."

Our takeaway: Both sides can be right at the same time. AI can genuinely be a great diagnostic tool and still be poorly governed when insurers use it to manage costs.

That's the real story here, not "who won the argument," but "who's checking the AI once it reaches the claims desk."