Re:An Ethical AI Never Says “I”
by jamilton on Sunday March 26, 2023 @06:13PM
(#35320191)
Un-"limited" LLMs (to me meaning pre-RLHF) look more like GPT-3, which instead of responding to a prompt will just continue it. As an example, you'll ask it a question, and it will give 5 more questions. To get it to write about something, you have to write it's introductory sentence.
[ Parent
| on HN ]
There are no comments on this comment yet.