The Font Causes the Font and the Font Does Cause the Font

I think there's just very little intuition about these algorithms like so if you feed an image into one of these algorithms, right? Like technically speaking anything in that image is fair game. And because of that pixel we can predict well, but we have no idea what that pixel means or we think that or that pixel is meaningless, right? Right I mean these examples though to me are glaring examples of like you didn't have a clinician informing what went into that model. You wouldn't include age in your model and so the one that was predicting kids If it had adjusted for age it wouldn't have done that. The lying down one  is a little bit trickier because It depends