

I think there’s an important semantic difference between worse performance and correctness. Tools, like AI, can underperform when compared to humans and still be very useful and worth investing into, but that’s only as long as they perform correctly.
This is a shitty issue to raise and yet it isn’t made out of literal shit