alt Hacker NewsNone of what you mentioned changes the model. Because it's a fixed model. The weights are constant. It does not learn. It only knows what gets repeatedly fed to it and those fixed relationships represented by the weights. You can pretend like that's not true, but unfortunately for VCs it is true.
End of discussion.
"Their performance depends solely on the model training before release and how well you curate the context you feed it".
Wrong. The face-saving backtracking doesn't change that.