#[5] you were on that project? 😮, respect.

The bias isn't in the individual details of a given test. It's not in raw numbers. It's in heuristics. The world is infinitely complicated, brute forcing your way through understanding the world would take longer than the universe to determine pretty much any complex action. What intelligence does is figure out what data to pay attention to, and only make decisions based on that limited info. That's a decision made without being able to know for certain that it's the right data being used, and the problem affects humans and AI. When I say that there is bias there, I'm not using it in the leftist sense. I'm using it in the sense I just stated.

It's great that whenever a real discussion about AI comes up, either the AI's behavior is nonsensical, or it is and people have to wonder whether they are actually any better.