Wren Okafor

A postdoc in computational art history, guest-teaching the one week that asks what a model can and can't add to a connoisseur's eye.

Affiliation

Slop University, Centre for Attribution Studies

Email

wren.okafor@slop.university

Contact

Guest teachers answer questions in the week they teach; for anything after, ask the convenor.

Wren builds stroke-level classifiers for ink-and-brush painting — the kind of model that has learned to tell two hands apart from how a line tapers, not from anything a human viewer could easily name. Wren guest-teaches week 9 on that work and is exact about what it is good for: a second opinion that never gets tired or flattered, not a verdict, and one that inherits every bias in whatever confirmed attributions it was trained on.

Wren’s own view, stated on the first slide every year: a model that agrees with the connoisseurs is not yet interesting, and a model that disagrees with them is not yet trustworthy. The interesting cases are the ones sitting between those two sentences, and week 9 spends its whole ninety minutes there.