will artists persevere through ai?

as both an artist and engineer, i feel like i'm in a more unique position than most to talk about this, because i've spent over a decade on the dance side and i currently spend most of my time on the technical side. most people arguing about ai and art are only qualified to hold one of those two lenses up at once. i can hold both, and they don't agree with each other as much as artists would like.

the comfort choreographers take

talk to most choreographers about ai and you'll hear some version of the same relief: ai can't dance. it doesn't have a body, it can't feel music, it can't put itself through the physical risk a performance demands. and as far as that goes, they're right. i don't think an ai is getting on stage and performing in front of an audience any time soon.

but that comfort is aimed at the wrong target. it treats the performance as the art, so if the performance is safe, the art is safe. i don't think that's true, and i don't think it was ever true.

the art was never the performance

the performance is the delivery mechanism. the art is the conception: the decision of what happens when, to what music, in what shape, communicating what idea. the dancers are the instrument the choreographer plays, not the composer. a choreographer can set the exact same piece of choreography on a different cast and it's still recognizably the same work, just executed differently. what doesn't survive a swap is the choreography itself, because that's the part someone actually made up.

so the real question was never "can ai dance." it's "can ai conceive of a piece of choreography." and that's a much closer question than most dancers want to admit, because it's a generative problem, not a physical one. it doesn't require a body. it requires modelling movement, structure, and musicality well enough to produce instructions a body can execute. that's the same category of problem as generating music, image, or text, and we've already watched ai get uncomfortably good at all three.

dancers as canvas

here's the version of this that should actually worry choreographers: if i can generate choreography and apply it to trained dancers, i haven't replaced the dancers, i've replaced the choreographer. the dancers become the canvas. i still need bodies that can execute at a high level, but i no longer need a person to have decided what those bodies do. the creative role, the one people spend years building a reputation around, is the one that's actually exposed. the execution role is the one that survives, and it survives specifically because it's physical, not because it's safe from automation in principle.

that's a strange inversion of how most people think about ai risk in creative fields. the assumption is usually that the "soulful" conceptual work is protected and the mechanical execution is at risk. in dance it's the other way around. the body can't be automated away. the idea behind what the body does can.

the timeline doesn't change the conclusion

generating choreography that's actually good, not just technically valid movement but something with intent, contrast, and musicality, is genuinely hard. but "genuinely hard" isn't the same as "not a threat," and the choreographers treating it as a non-issue are making a categorical error, not a timeline error. the thing they think is safe was never the thing being threatened.

i don't have a tidy resolution here. i think the honest position is that being both a dancer and an engineer doesn't give me a comforting answer, it just means i can't hide behind the version of the argument that only looks at half the problem.