@cjd @Moon
Obvious jokes aside, the problem is, that you cannot create a dataset by humans, since humans are incapable of making this distinction themselves.
The entire concept of schizophrenia and inteligence being 2 sides of the same coin does apply here 10fold. Because, briliant people do see paterns that you cannot visualize, that means, that you cannot know, if they actually are smart, or if they are bullshitters.
This is, why most atempts at doing this end up just with recognizing, how niche words you use, since the niche words are needed to make a scientific article. But, you immediately turn to social science loons, who cannot form a single sentence without going full systemic prejudice against margenalized methaphors for cheese.
@cjd @Moon
And to finish this, take conspiracy theories.
They are theories. By definition, they are something, that people who believe in them weren't able to disprove.
But, when you talk to a normie, they take the word theory, and believe, that it means by definition, that it has to be false. This is the inteligence on the 100IQ level. But, it is based on a deeper problem. That is, that we are working with an insanely limited amount of measurable data. And AI can only compate concepts to concepts, but It cannot compare concepts to data. That means, that it has limit to it's potential only to things, that humans have already measured.