(LLM Tests) I still cannot find a competitor to GPT4 in the ability to conduct r

(LLM Tests)
I still cannot find a competitor to GPT4 in the ability to conduct reasoning. All of them are bad. GPT is simply the least bad of them.

I really do like the Dolphin uncensored as the future, mixtral is getting there, I use perplexity for research, and GPT4 for assistance with search and writing.

I suppose if I wrote programming code (a via positiva) instead of legal pseudocode (a via negativa) then I might feel differently.

So far, the AI’s are sh*t at predicting fields of possibilities, subjecting them to competition, and then using falsification, so that from that result they can reason.