G3pgxX9WEAAegjn.jpeg

Hahaha. We're doomed.

Oh, I'm dying! Probably literally if this is who welcomes me to the hospital one day.

I mean yes, we're all doomed. But hasn't this literally happened with human doctors?

I am not sure, a drunk surgeon maybe? But what you are saying translates to > "it is ok to let a machine out there that could kill any time, since humans kill sometimes too"

I mean I could tell ChatGPT it's wrong and 2 out of 5 times it'll say "You're absolutely right" even when it was correct the first time.

Human beings: make mistakes, sometimes illogical, non-deterministic. AIs: make mistakes, sometime illogical, non-deterministic. I'm not saying they're the same thing at all, or that we should use AI in all or even most applications, and I most certainly didn't make any moral judgements or ever say that it's okay to kill anyone under any circumstances as mc seems to imply (I literally just asked a question). I do however think it's very reasonable to point out that human being also make mistakes, and there is an inherent risk there, which needs to be carefully weighed against any proposed alternatives (whether AI or not). This cartoon certainly makes a valid point in a humorous way that AI is likely being over-applied in many contexts, but it's humorous because it's highly unlikely that anyone in the real world is seriously proposing letting AI perform surgery completely unassisted right now. But there are lots of medical applications where AI is hopefully going to help. For example, there's some evidence that AI is actually better than humans at reviewing PET/CAT/x-ray scans for certain anomalies and diseases, and if that's the case, I for sure want them to use that along with a qualified oncologist or whatever. I think predicting the future is extremely difficult, and it's real hard to say right now whether we're at the dawn of a new era or at the peak of a hype cycle (or maybe both like the .com boom/crash in the late 90s/early 00s).

I don't think it is going away any time soon, or maybe ever. And I agree with what you said. I just saw many people make that argument. It feels like they want to say "Look, I am not claiming that I am perfect as a human being, and you are not too, so don't treat AI like that."... and that should be an argument to unleash it on everything maybe.

I'm a solid member of the LLM skeptic camp. I see both promise and problems with the tech. There's a lot of over-reliance on them by people, and it could definitely get them into trouble. But I am definitely looking for more uses for them - every time I find one, it's another chip off of the skeptic rock. Skeptical <> hatred/dismissal/rejection. I am currently negative on the way the market and investment worlds are viewing it without looking at the larger pictures, and I believe there are way too many promises that will never be reached by reality (on the current path), but it's a fascinating technology with pros and cons. Like every other big technology. I've told my users to think of it like a tool on their bench. Screwdrivers have many applications: they tighten, they loosen, they pry things open, and they can even get wrapped in a rag to get all that paint out of the rim before you seal the can. But use them without care and they can go through your hand.

I am currently negative on the way the market and investment worlds are viewing it without looking at the larger pictures

I don't mean to topup what you are saying just for the sake of it... but modernity as a whole is like that. Market expansion usually ruins everything in the way.

True. I actually like being wrong on tech, markets, etc. It's fun to see something impressive enough to overcome my expectations. There are just enough ripples on the water right now that make me nervous about the fallout if/when there's a correction (or worse a bubble collapse).

Do you think a better technology could emerge?