utzer :verify: @utzer@soc.utzer.de
@denschub@mastodon.schub.social I always wondered why OpenAI is not using a better base prompt, because many problems can be caught by that. But yes, AI is a tool that is not reliable and I would not suggest using it if you don't know the topic or even don't want to check the output.
Dennis Schubert@denschub@mastodon.schub.social
@utzer@soc.utzer.de or maybe, just maybe - hear me out - a "please don't lie to me" prompt actually doesn't do anything to prevent AI from lying to you, because if it did, OpenAI would have already done it.
wild concept, I know.