bad point well made

Machine-generated art and the envelope of taste

This is a fairly well-trodden subject, but it’s one I wanted to gather some strong (and not so strong) feelings about. A blend of second-hand takes.

First order, linear thinking dictates: if I love this particular style of art (used in the general sense of: art, music, literature and digital creative media in general) and I am shown more of it, I will experience more pleasure. However, we instinctively know that the 1000th listen of our favourite track probably doesn’t quite hit the same as the first few listens. Experiencing art physically changes us, each subsequent exposure to the art will prompt a different response in us, however slight. The art and us (and importantly, how we feel about the art, our taste) exist in a dynamic system with feedback, with higher-order effects.

There’s a view (often implicitly held) right now that runs something like: if AI can so dramatically lower the financial/time barrier to generating what is currently perceived as high-quality art, then we can assume that the creators will be out of a job and the audience will be swamped with an infinite supply of highly stimulating content. This is the first order theory.

We know this rarely plays out in practice. I suppose as humans we have sensibly evolved with mechanisms to prevent us from deriving infinite pleasure from things, although there are examples of where these mechanisms have somewhat broken down (the infinite scroll, eating disorders). Commoditisation happens, we generally get bored, and we move on. For artistic innovators, this process probably happens quicker. Innovation is often stimulated by this commoditisation/boredom cycle.

David Foster Wallace put it well in an interview when talking about literary movements:

But after the pioneers always come the crank turners, the little gray people who take the machines others have built and just turn the crank, and little pellets of metafiction come out the other end... Academia and commercial culture have somehow become these gigantic mechanisms of commodification that drain the weight and color out of even the most radical new advances. It’s a surreal inversion of the death-by-neglect that used to kill off prescient art. Now prescient art suffers death by acceptance. We love things to death, now. Then we retire to the Hamptons.

We could liken the rise of machine-generated art to that of clip art. Before CD-ROMS and the clip art wars of the 90s, there was an industry dedicated to creating art of a similar aesthetic for a similar purpose. Clip art quickly became commoditised - putting it in your PowerPoint presentation soon became tired, and people moved on. There’s an irony that the forefront of art devaluation (inevitably) remains the humble presentation, as people now liven up their bullet points with a healthy accompaniment of slop.

The difference now (as it seems with many things in our particular stage of the Anthropocene) is that the change is happening much faster and much more broadly (in relation to existing artistic styles). Slop is moving up the value chain surely faster than any other artistic “style” ever has, if we can call such a nebulous collection of existing styles a style in itself…

A truly interesting question to be asking right now is how this speed and breadth of change in the economics of art creation will interact with our envelope of taste and therefore artistic innovation.

A pessimistic view would suggest that the honeymoon period of novelty from new artistic movements that had been enjoyed in the past will shrink to almost nothing. This is the age of near instant commoditisation, where DFW’s “little gray people” immediately arrive, turning the crank, draining all the weight and colour out of artistic innovations. Perhaps the sheer frenzy of this process is exciting in and of itself, but I’m not sure.

The optimist in me hopes that, as before, the relentless march of our envelope of taste shoves us into brand new spaces that we can hardly really conceive of right now. Let’s not forget, the machines aren’t just driving us away from where we are idling, but also potentially giving us the tools to explore further.

Equally, both of these eventualities might also sound equally dystopian. There’s a feeling that perhaps we aren’t well adapted to such relentless change.

Part of me wonders whether there will just be a dramatic shift in the kinds of artistic innovation that we see. If the “value” of art is (somewhat) linked to its scarcity, then perhaps the kind of highly valued art whose scarcity was largely financially driven will fall away. Just as photography dramatically devalued the portrait artist, perhaps certain (not all) films that required huge crews to film will similarly.

What I find interesting is that the art whose scarcity and scarcity-linked value is not due to financial constraints but more due to niche appeal, ie fringe art, might be less affected by this commoditisation. Yes, a machine could generate this very easily, but why would it (amongst the sea of all possible other options)? Perhaps the future of artistic innovation is a fragmentation into the niche. Maybe it always has been.

Another interesting question is to what extent are the necessary ingredients for taste (that is, the dynamic quality that allows us to create something that is enjoyable to someone) available to the machines? Maybe the answer is already here. Either way, it feels like the machines might be taking us closer to some deeper kernel of what we value in art.

Further threads: