Re:Weird GPT-4 behavior for the specific string “ davidjl”
by waboremo on Thursday June 08, 2023 @01:16PM
(#36245335)
It isn't, yet it's used that way.
[ Parent
| on HN ]
Re:Weird GPT-4 behavior for the specific string “ davidjl”
by krapp on Thursday June 08, 2023 @02:36PM
(#36246548)
Everyone is worried about doomsday scenarios like runaway AGI and Skynet, meanwhile the real danger of AI turns out to be people naively assuming it works like the computer in Star Trek, that because it can hold a conversation, is has some form of intelligence and awareness.
Re:Weird GPT-4 behavior for the specific string “ davidjl”
by JasonFruit on Thursday June 08, 2023 @03:24PM
(#36247301)
Please keep saying this over and over. Even people who get that it doesn't work that way don't seem to grasp the danger of others relying on these tools in inappropriate ways.
Re:Weird GPT-4 behavior for the specific string “ davidjl”
by krapp on Thursday June 08, 2023 @08:08PM
(#36251196)
The problem is LLMs are kind of an outside context problem for our culture. We're still arguing over the correct language to even describe what it does ("hallucinate"? "confabulate"? "lie?") because every word we have presupposes the existence of a stateful mind with intent. It's a philosophical zombie, which is so counterintuitive most people can't even accept it as a sensible concept.
Between the hype and decades of indoctrination through science fiction about how AI is "supposed" to work, I guess it isn't surprising that this is how things are shaking out. People will learn, the way they learned with the internet, but it will take a few years I think.