←back to thread

Microsoft and OpenAI's close partnership shows signs of fraying

(www.nytimes.com)
321 points jhunter1016 | 1 comments | 18 Oct 24 11:11 UTC | HN request time: 0s | source
Show context
twoodfin ◴[18 Oct 24 12:08 UTC] No.41878632[source]▶
Stay for the end and the hilarious idea that OpenAI’s board could declare one day that they’ve created AGI simply to weasel out of their contract with Microsoft.
replies(4): >>41878980 #>>41878982 #>>41880653 #>>41880775 #
candiddevmike ◴[18 Oct 24 12:53 UTC] No.41878982[source]▶
Ask a typical "everyday joe" and they'll probably tell you they already did due to how ChatGPT has been reported and hyped. I've spoken with/helped quite a few older folks who are terrified that ChatGPT in its current form is going to kill them.
replies(5): >>41879058 #>>41879151 #>>41880771 #>>41881072 #>>41881131 #
ilrwbwrkhv ◴[18 Oct 24 13:03 UTC] No.41879058[source]▶
It's crazy to me that anybody thinks that these models will end up with AGI. AGI is such a different concept from what is happening right now which is pure probabilistic sampling of words that anybody with a half a brain who doesn't drink the Kool-Aid can easily identify.

I remember all the hype open ai had done before the release of chat GPT-2 or something where they were so afraid, ooh so afraid to release this stuff and now it's a non-issue. it's all just marketing gimmicks.

replies(7): >>41879115 #>>41880616 #>>41880738 #>>41880753 #>>41880843 #>>41881009 #>>41881023 #
usaar333 ◴[18 Oct 24 15:58 UTC] No.41880616[source]▶
Something that actually could predict the next token 100% correctly would be omniscient.

So I hardly see why this is inherently crazy. At most I think it might not be scalable.

replies(5): >>41880785 #>>41880817 #>>41880825 #>>41881319 #>>41884267 #
1. kbrkbr ◴[18 Oct 24 23:01 UTC] No.41884267{3}[source]▶
That does not seem to be true.

Either the next tokens can include "this question can't be answered", "I don't know" and the likes, in which case there is no omniscience.

Or the next tokens must contain answers that do not go on the meta level, but only pick one of the potential direct answers to a question. Then the halting problem will prevent finite time omniscience (which is, from the perspective of finite beings all omniscience).