So LLMs can basically describe any image, there studies where they're better than the average GP at medical diagnosis.. Caveating all the awful tech bro stuff, in terms of accessibility tech they're revolutionary, I think they should be used and developed responsibly - what am I missing?