Yeah that was my point exactly.

What they should be worried about is the training dataset. And by that I mean the people and its actually-expressed opinions. All they’re saying is that “most people are woke”, basically, which I take as a win. Not for AI but for people.