> There is, of course, no way of knowing how accurate Dewey-Hagborg’s sculptures are—since the samples are from anonymous individuals

From the second one. So who knows, it is fairly likely to just be a guess? If it were real, I'd expect better uses of it than just by bored artists.

Edit: I should have checked the first one, they at least show an experiment (if a sloppy one). The results are...not great it seems if the goal is it being recognizable as the person in question.

Would love to see representative pairs of predicted vs real faces.