[epidemian]

Yeah, i agree that, on principle, generative AI should be a good got for these kinds of things. But, the results are... idk, uncanny. At least to me.

For example, if you were to ask a human artist to draw this image with two crabs, and this artist didn't know how many legs crabs have, what would they do? I think they'd probably google how many legs crabs have, and how they look in general, and at the very least draw two crabs with the same amount of legs.

But these generative models don't know how many legs crabs have, and they are not clever enough to ask, or even count it seems. So you end up with results like these. The main crab has 8 legs (6 + 2 claws), while the little one seemingly has 10 (or at least it has 5 on one side), and it's also missing an eye for some reason.

A human artist can also make mistakes, for sure. But i don't think these kind of mistakes would be expected from an artist.

And i agree with the other commenters that said this reflects poorly on the author of the book. Didn't they care to check the cover for obvious mistakes? Attention to detail is a very important trait on a technical writer.