Existential ComicsMar 3
Always a good time these days to repost this slide from an IBM internal presentation in 1979.
Show threadExistential Comics
In other words, maybe...and I'm just brainstorming here, it is actually BAD for an AI Chatbot to decide who to bomb, etc.
Mar 3 at 4:42pmWeb
Show threadɹ uɐp Mar 11
@existentialcomics I am the language model of a modern major general
Show threadBenAvelingMar 11
The AI didn't 'decide'. The AI printed some random text that has the format of a decision. It was humans that decided to treat that random text as a substitute for intelligence.
@existentialcomics
Show threadtuban_muzuruMar 11

@BenAveling @existentialcomics

The TL;DR for any LLM is someone practicing their tennis game by bouncing the ball off the wall. That's kinda it:

Appropriate . But neither Right nor Wrong. If the AI Chatbot was asked to decide - who asked the question?