nxpnsv 8 hours ago

I guess that the point is that doing so already is not safe?

bigstrat2003 2 hours ago

You already shouldn't be using LLMs for either of those things. Doing so is tremendously foolish with how stupid and unreliable the models are.

awkward 8 hours ago

There are several humans who need to make decisions between bad training data and life or death decisions coming from an LLM.