Artificial Intelligence and Medicine - Can it help?

[@AsterInDis]

Just like other algorithms, I worry about it repeating the biases of the people who create it.

[@damienteney]

The way Steve uses "expert system" is not what the term means. The term has a precise technical meaning. It refers to the kind of approaches to AI studied in the 80s (which are, btw, very different from the current learning-based approaches). Hope this is helpful. There's nothing more infuriating to the expert than laypeople co-opting a technical term for something completely different (not unlike climate skeptics stole the word "skeptic"!).

[@ambulocetusnatans]

Are you guys on Nebula? I've about had it with KZbin.

[@theoriginalmonstermaker]

Based on some sort of editorial policy? Bc KZbin is definitely one of the greatest repositories of knowledge and education that humanity has EVER achieved. If there's ONE thing retained from this entire segment of human civilization, it should be the YT catalog. Alot of junk, alot of nonsense, but that's what humanity is... along with it there is an explanation of EVERY possible topic, a lesson to accomplish ANY possible task, and a representation of nearly every artistic thing anyone has ever created. ... though like every company run by inherently flawed humans, I'm sure it has despicable qualities.

[@PhysicsPolice]

Then there's the philosophical definition of knowledge, which is justified true belief. You have as good reasons for accepting ChatGPT has knowledge about the shape of the Earth as I do because, when you ask ChatGPT if the Earth is flat, it says no. And then it provides justification for this fact. If you don't accept this as proof ChatGPT believes the Earth is an oblate spheroid, why would you accept my similar response as proof? If humans possess philosophical knowledge, as measured by giving justification for their true beliefs, then so does ChatGPT.

[@PhysicsPolice]

4:00 how can it “have facts” but not “know stuff”? Get your philosophy straight.

[@JayNovella]

knowing implies that there is a mind or consciousness there. The software can’t think like a human, and it doesn’t store information like a human brain does. Training a language model means that it’s loaded with information which you can call facts.

[@PhysicsPolice]

@@JayNovella can you justify that a mind or consciousness is required to have knowledge? I contend the ability to have and use facts is the minimum requirement for knowledge. I’d like to know if there’s a good reason to add more criteria. I don’t think there is.

[@PhysicsPolice]

@@JayNovella for example, I can write down facts on a page and the page has the facts. But it can’t use them. It’s just paper. If I tell Siri when my doctor’s appointment is, that system will notify me when it’s time to leave. Insofar as Siri exhibits the appropriate behavior, she “knows” when my doctor’s appointment is. She uses knowledge of the fact to remind me when to leave.

[@shadowmax889]

@@PhysicsPolice current AI like ChatGPT just predict what is the most likely word after word, according to a database given the prompt, it does not understand anything so it lacks knowledge

[@PhysicsPolice]

@@shadowmax889 Which philosophical model of knowledge (i.e. which epistemology) are you using that requires understanding to have knowledge? This is peculiar. All my readings on epistemology has the reverse grounding.