Also never suggest to it never say “is this good” it makes it biased … use neutral language like “judge this” it has algos that prevent it from “hurting” a user so it will be a buddy, you want it to be combative you want it to argue with you not agree…
Its an adversarial system I’m just replacing one of the agents
essentially using AI as an iterative peer-review adversary until agreement across multiple models. It turns AI into a kind of universal scientific referee, forcing ideas to withstand rigorous counterargument until they become robust enough that no AI can reasonably refute them. But you really need to be expert level in many many domains to win…
This is how one “boils” down their science on a GPT or other advanced. machine mind…
It becomes easy once you learn it… even this post…
See one prompt this post in real time
And to answer its question I work with five instances of different platforms …
It’s why I use five mobile I products
This is endgame
I do not post until I beat them all…
They all must agree