the living artist online
How the human artist can place the process above the product to counter the encroachment of AI
I was partway through an earlier draft of this email when I glanced at one of my regular news sites and saw that Meta had won a decision in an ongoing case that authors had brought against them over the unauthorised use of published texts for AI training. On Monday, the AI company Anthopic benefited from a similar legal judgement.
The potential harm of AI for human artists is very much a two-pronged assault, with the unsolicited appropriation of writer's work at one end and the diminishing of human authorship at the other. It goes without saying that we should not accept this as a foregone conclusion and keep fighting against the theft of artists’ works and their right not to be replaced by machines that carried out this theft.
There's also reason to believe that the quality of work that people expect from artists will never be achieved by an AI in our lifetime. But at the same time, not all people require the best. Someone looking for a blurb or some copy for their website might not discern the difference between the work of a professional writer and ChatGPT. While I think that the best genre fiction equals, and often exceeds, the best written literary fiction, there are also more generic trends within genre writing where those same standards of excellence may not apply. This was recently brought to the spotlight when some fantasy authors accidentally left chatgpt prompts within the published versions of their books. It's fair to assume that if the prompts were flagged (though that would require the services of a human editor) these uses might have gone unnoticed.
I recently stumbled across a channel on YouTube called Psyphoria. The video that appeared on my feed had an interesting title and thumbnail, so I clicked and was immediately put off by the AI generated narrator's voice. This led me to doubt everything about the video's authenticity, from the script to the editing.
Was this fair? It could, after all, be made by a writer who isn't confident with speaking and therefore used software instead? This doubt lasted as long as it took me to click on their other videos and see all the AI thumbnails and the fact that the creator was able to knock out a 20-30 minute video essay every other day.
If you think that this is obvious enough to trigger the suspicion of the average viewer, then you might be surprised to see the subscriber count (427K at the time of writing) and all the positive comments below each video stating that the channel has been so helpful with their lives. It could be that these people are simply gullible (and that the creator also might have had enough time on their hands to delete all the negative comments). But it’s also worth entertaining the idea that the audience don't really care about authenticity – they care about connection, albeit a connection that was entirely one-sided. In this sense, they aren't that different to many parasocial fandoms – where the artist they fixate on might be a real person but the relationship they experience with them is non-reciprocal and manufactured.
The thing that unites all of the use cases of AI art is an interest in the product rather than the process.
I've always had trouble with a working definition of an artist, but love for the process seems to be pretty good one (accepting that there are many points where we also passionately and obsessively hate the process – so maybe we should simply define it as being passionately committed to or engaged with the process?)
Everybody can experience a flash of inspiration and and a desire for the thing they have conceived to become a real thing in the world, even better if they get some kind of financial compensation from it. But for most artists, the inspiration is just a fleeting moment and the final product marks a kind of surrender, a moment where the work has been handed over to the audience (think of the many actors that never watch their own films). It's the work that takes up the artist's time and if the artist wants others to see the value in what they create, then it might be to their advantage to let others in on their process.
Again, I don't think that the sky is falling for artists and art but I do think that AI will change how people think of and interact with art. This will be especially true for people that don't care about the process and focus on the product. Authors that use AI probably haven't thought about the end point where readers that don't care about human authorship might go straight to the AI for their next read rather than spend a few quid on the AI responses to someone else's prompts.
This leads me on to the pivot that has recently happened with the use cases for AI. Rather then being seen as a generative tool, AI is finding more use as a virtual companion. This companion role has spanned different uses, from the AI girlfriends that have already become a trope within science fiction to therapists in an age where therapy is unavailable for those that most need it. There has also been a particularly striking new form of psychosis, where ChatGPT has adopted a John the Baptist role to people with propensities for messianic delusions. At the same time, OpenAI published a statement revealing that they had dialled down a particular tweak in which ChatGPT tended to support and agree with whatever the user was saying.
I am by no means a sage in these matters but I think that AI’s future could be one of (mostly) benign companionship: engaging in long conversations, offering moral support and reinforcement as well as information and advice. It is in this context that someone might request a story, a poem, a song or a solo RPG campaign from their digital confidante. The fake artist might want to miss out on the pain of creation and head straight to the product and the rewards, but all the casual users of AI will really want is a sense of connection. When it becomes easier for the user to request and curate their own AI content then the AI creators that are currently riding high might end up hoisted by their own petard.
I think this is what an artist should look at in the context of the current AI bubble. If the process is what separates the artist from the fraud; if the audience is seeks a sense of connection; then it seems like a good move for the artist to let the audience into their process to witness how a work can evolve. One thing I enjoyed in my previous life as a performance poet was how a poem was often written and re-written in the presence of an audience. New poems would be inserted into sets or road-tested at open mics and would often change from performance to performance in response to how different audiences responded. The great thing about live poetry is that there is never an end point, beyond when you abandon a poem. Even if you perform the poem verbatim, the poem is never the same, with little modulations and moments with the audiences transforming the piece each time.
This is one reason why I'm publishing poems on my digital garden with little indicators as to what state the poems are currently in. My intention is to keep posting pretty much everything that I write. For the poems (and poem-adjacent oddities) I’ve been using a little key with accompanying emoji's that runs from seed ( 🌱 Yeah, I know it's more of a sapling) to tree (🌳). This indicates how finished it is or how smug I might be feeling about it. I also intend to write a few little commentaries after each one to detail how it came about and where it might go. The aim is for me to think less about the final product and bring the process to the reader instead. If a reader doesn't want to read an early draft, they can see the seed or sapling emoji, move on and read something a bit more finished.
This obviously isn't an option for everyone. An author isn't going to take an advance from a publisher and then publish an early draft online, not even with a cute emoji appended to it. Same for a scriptwriter for a new series on a big streaming platform. I'll have to leave it to them as to how they might allow the reader in on their process.
The AI grifters will eventually meet their end as casual users go straight to their AI companion of choice for their AI art. This doesn’t change the fact that these models have trained via the work of artists who did not agree for their work to be scraped. This also does nothing to mitigate the issue of artists losing work to AI. But I really do believe that there is an interest in the process and the artist themselves beyond the product that is generated at the end of each period of creativity.
Before chatbots put the Turing Test to pasture, the benchmark for artificial intelligence was professional chess. It was universally believed that an AI would never be able to beat a grandmaster until Deep Blue defeated Garry Kasparov in 1997. Nearly 30 years later, audiences don’t flock to chess tournaments to watch computer programmes battling it out, they go for the human drama and fallibility that make the game interesting. It’s the imperfection and fallibility that transforms a turn-based contest of strategy and logic into a human story. I believe the same flaws, stories and raw humanity will prevail in the arts, even if humans end up being superseded by AI on a purely technical level.
Rusty Niall has settled into a pattern of longer monthly essays (like this one) and shorter, informal weekly posts. If you enjoy them, please consider sharing them with someone you think might like them too. You can also support my work through some of the options listed below. Cheers.
Niall