> trust as a counterweight to probabilistic answers
The basic problem with that is that even with 5 9s of good results, that remainder still gets you very public cases of "put glue on your pizza". And we're not even at 5 9s yet.
The basic problem with that is that even with 5 9s of good results, that remainder still gets you very public cases of "put glue on your pizza". And we're not even at 5 9s yet.
When does the LLM think that will happen, please ask it I'm curious
Nothing about filming food commercials. Why white knight an undercooked product?