@ShaulaEvans @bookstodon

Far too much depends on the originality and voice of the artists for AI to ever make much of an inroad.

Voice is the one thing the technology is structurally bad at. A model trained on everyone’s drawings converges on the center of the distribution, and what makes an artist worth reading is how far they sit from that center, and why. A wobbly line, a strange way of cutting between panels, a recurring obsession with some corner of the world: these are accumulated decisions made by a particular person with a particular history, and they carry meaning because they were chosen.

There’s a second layer: readers of this form tend to know the artist. Part of what they’re buying is the relationship, the sense that someone made this with their own hand and their own stakes. Take that away and you’ve got something closer to stock photography, which is a fine business but a different one.