OpenAI API
Jun 07, 2023
This is a best-practices guide for GPT. It's something we could probably have used last fall, before any of this hit the internet. It might have tempered expectations somewhat. For example, this should have bee posted very clearly everywhere GPT is used (including chatGPT): " GPTs can confidently invent fake answers, especially when asked about esoteric topics or for citations and URLs." *sigh* See also this Hacker News thread on the best-practices guide.
Today: 7 Total: 162 [Direct link] [Share]
