@cmccullough Predictive AI helps detect cancer and negate human error if trained properly btw
They are being trained on cancer samples from patients and other conditions
LLMs that dont do image gen or are ethical like Claude are fine
Saying "stop using ai" is like saying "stop using fire because its the root cause of forest fires"
@jarinks @cmccullough I'm not sold on Claude being a golden boy here. I've gotta do more research before I'd say so. I think the resource strain and entanglement with Facist governments is probably as bad as Google or OpenAI but again, don't know.
All I can say is the most ethical of LLMs are the least useful to me. Please cure cancer if they can
@jarinks @cmccullough if I could be as confidently wrong, I’d be a billionaire.
Specialized models are older than AI bubble and don’t consume enormous amount of resources in training and operations.
Ethical oracle models, however, don’t exist. They require climate destroying amount of resources for training and operations alike by their nature. Claude is no exception here.
If you're telling me we should leave aside the ethical concerns for the moment, then you *know* that what you're advocating is terrible