I just saw a screenshot of someone pasting I think the US bill of rights to one of those AI write detectors; the site concluded that the text was written by AI.

> How much of our opinions are driven by context

I’d say for art, a lot? There’s a ton of art that a halfway decent painter could do now, the art of it was being the one to do it originally. At least that’s how I, as an absolute philistine in that regard, understand it ;)

There is a similar experiment where a famous violinist plays in a subway station. Nobody really notices or appreciates him and his music. https://m.youtube.com/watch?v=hnOPu0_YWhw&ra=m

Yeah I agree, in art a lot is driven by context: there's so many paintings or songs that are not outstanding in itself, but the full human context around it makes it significant.

That brings up the idea that art can be "outstanding in itself", aesthetic in a vacuum, disconnected from what people are caring about. That's dubious, but anyway the AI art doesn't attempt that. Instead it has access to a lot of freeze-dried human context which it rehydrates and presents like a fresh meal, so it partially succeeds at providing that significance.

You're right. Maybe I should have said 'painting or songs that do not SEEM outstanding in itself". My point is that an AI 'rehydrating' human context that you mentioned, is (usually) not enough to get the same significance as human-made art.

At least, for now.

For an edge case: people will be impressed and interested if you tell them that a piece was painted by an elephant, and then suddenly unimpressed if you tell them you were lying about that. So one function of art is as a sort of experiment, like the art is experimental data, where authenticity matters, because the interest is in the demonstration of a perspective, the reactions of an artist in the situation. Consider noir: a movie is much more plausibly authentic noir if it was made before about 1963, that is, if it was made by actors and directors who actually wore those hats (and lived through other tropes). Later on, it's imitation, regardless of how accurate: the experimental data is invalidated, it doesn't (seem to) mean so much.

This feels like the example of (world-famous violinist) Joshua Bell playing violin in the DC subway and getting just a few bucks. It's totally different than paying money to see him in a concert hall, context matters so much...