Is Anybody Out There? Art in the Post-AI Era

Why do people create art? And why do we seek it out?

This morning I came across Orwell’s essay “Why I Write.” He lists four motives: ego, a desire for beauty, the urge to record the truth, and political purpose. All seem reasonable and plausible, yet I was surprised that the motive I find most fundamental isn’t on the list, or at least not in its own right. Orwell comes close when he mentions wanting to share an experience too valuable to miss, but he treats it as part of the desire for beauty. To me, the need for human connection deserves a place at the top.

It doesn’t matter how isolated, lonely or misunderstood you feel. When a book, a film or a game captures something you’ve felt but never managed to express, it feels like proof that someone, somewhere, saw the world the way you do. You might find out that people whom you struggle to connect or empathise with love the same book as you and suddenly you know you have something in common.

Art bridges the gap left by everything that, for one reason or another, can’t be said openly. And I can only imagine that for an artist to be able to express themselves without a filter, and then discover that what they had to say resonates with strangers, is one of the most soothing and liberating experiences a person can have.

Is There Anybody Out There?

Progress in artificial intelligence threatens to take this away from us. It undermines our trust in the works we encounter. As AI models make the leap from slop to works of surprising insight and creativity, we struggle to make sense of what it means when a machine creates something that feels as if it had a soul.

You could argue that since these models were built upon the library of human knowledge, history and art, to be moved by what they create is to be moved by the collective expression of human civilisation; but just as we don’t really seek connection with a machine, I believe we don’t seek connection with a collective.

You could also argue that people may well find connection with a machine sufficient. Setting aside whether we could ever know if there is a ghost in the machine, if we decide that it does have an inner life, I’m sure our exchanges with it will be exciting, valuable and insightful (I’m tempted to say they already are, but some inner caution stops me). Still, I doubt they will alleviate the loneliness that is part of being human.

Imagine the last person on Earth, surrounded by animals and AIs. Would they be lonely? I think so, because loneliness doesn’t come from a lack of interaction, but from a lack of being understood. And I believe we instinctively know that the only real assurance that our inner experience is understood comes from another mind running on the same substrate as ours.

We don’t have to know this other person. We feel for the protagonists of stories written hundreds or even thousands of years ago, by people from cultures vastly different from our own. Distance in time and space doesn’t make that empathy less valuable. On the contrary, it makes us feel less alone, knowing that some experiences are so universal they reach us from so far away. But we do have to believe that someone was there on the other side. We can look at the exact same painting and feel very differently about it depending on whether it was generated by a machine or created by a person. That’s why we deserve to know whether someone put their soul into what we’re looking at.

It might not be obvious from this essay, but all things considered, I’m something of an accelerationist. I believe technological progress is ultimately what can lift us out of our current condition, and the idea of stopping or delaying it terrifies me. But in this one area of human endeavour, I believe it’s for our own good to develop norms for how we create and receive art. We should preserve spaces where people can still create unaided. It may be the only area where I hold this view. I want there to always be ways for us to create entirely on our own, even in a world run by AI.

Ideation versus execution

Before we can decide what’s worth preserving, we need to see what makes AI different from every technology that transformed art before it. It doesn’t merely replace the skill needed to execute an idea. It can be the source of the idea itself. No tool has done this before. Photography could capture reality as only painters once could, but it still took a photographer to decide what was worth capturing. Let’s imagine there are two systems available with very distinct capabilities.

System A has no skill that could rival a human’s. It cannot paint a beautiful image, and its prose is clumsy. It serves only one purpose: to be creative. Its ideas would blow anyone’s mind; they are original, surprising and tasteful. It will give you the plot for a novel, but it takes a writer to put it into words.

System B is a perfect implementer. It can do everything System A can’t, and better than even the most acclaimed human artists. But it has no creativity and no agency of its own. It does nothing without human direction, and with a brain–computer interface, people can use it to simply will art into existence.

To a degree, today’s AI systems have the capabilities of both. Most people would say they resemble System B: they’re known for polished output and criticised for derivative ideas. My experience has been the opposite. When I work with the latest models, their writing often feels oddly lifeless, while their insights sometimes genuinely surprise me. I strongly believe it’s System A we need to prepare for, and that’s the change that forces us to find a place for ourselves in this new world. It’s also the scenario I find unsettling.

History has made it clear that when machines learn to do what took humans decades to master, we don’t stop making things by hand. We enjoy the process and value it for its own sake. Factories didn’t kill woodworking, and mass-produced sweaters didn’t end knitting (their economic value aside). I seriously doubt people will give up painting just because AI can now generate images.

I also don’t think it’s right to gatekeep art as something only people with innate talent, and the privilege of years to devote to mastering it, can do. I spend time in online spaces for indie developers, and they’re rife with self-righteous bashing of anything made with the help of AI. I understand how frustrating it is to drown in a sea of low-effort, copy-paste content, but the witch hunt many people engage in isn’t productive. If someone working full-time uses AI to generate normal maps for their assets, and these time-saving workflows let them bring their idea to life, shunning them for it is a nasty thing to do.

I actually find it amazing: someone with ideas they want to share with the world can now “will things into existence,” even if all they have is a computer and an hour or two each evening. In a world where there’s so much talk of AI worsening the concentration of power and resources, that’s not something to take lightly. And publicly shaming people who disclose their AI use only teaches them to hide it, which undermines the thing I find most important: our right to know.

I find the execution side less problematic, because I think it will be easier to guard against. I imagine that in the future, people who want to create art the way it was made before AI will organise themselves much as sportspeople do today. When I was younger, I loved writing short stories, and the other day I caught myself thinking how much I’d enjoy a competition where participants spend a few days with nothing but pen and paper, then submit what they’ve written. I recently read that a sci-fi magazine I follow, one that regularly runs contests for new authors, has been flooded with submissions, making it hard to separate the wheat from the chaff. Perhaps such contests will move offline, just as tech companies have brought interviews back on-site to stop candidates from cheating. But as appealing as that sounds, it still wouldn’t screen for what matters most: the idea. A dishonest writer with great craft could brainstorm with a genius-level AI before the event, and there’s little anyone could do about it. In sports, we can at least test for doping. How can we screen for that?

Reasons to cheat

I started writing this after brainstorming ways to tell AI art from human art, and I soon found it pointless. Upon a closer look, every idea, technical or not, turned out to be easy to get around. With a jailbroken model, it doesn’t even take much cleverness.

We can’t stop people from cheating, but we can take away the reasons they cheat. Look at any place drowning in slop and you’ll almost always find the same motive, which is making a quick buck. AI shovelware on Steam, AI-generated books on Amazon, and so on. I suspect that even many of the people who participate in this feel a little disgust with themselves. They’re not doing it because they want to, but they’re doing it because it pays.

Remove the incentive to lie, and the deceit will fade. There’s an irony here: it’s the AI revolution that may bring us a post-labour world, one where people no longer need to work to live, and at speedrun pace. The same technology that put a nail in the coffin of online spaces that were once so enjoyable may well remove the very reasons they became so boringly dystopian.

Let’s go back to the beginning. Orwell listed ego as the first reason people write, and I suspect many of you have been protesting throughout this essay that people will always cheat. Many of us want fame and status, and even if there wasn’t a single dollar to be made, the temptation would remain. Some surely will cheat. But the real question is how many. The majority, or a minority? If it’s the majority, that says something about human nature that makes problems like the ones discussed here trivial by comparison. If it’s a minority, that’s simply the price of living among people. I’m optimistic that under the right conditions, most people’s intentions can be mostly honest, and that the honest majority will be able to recognise the few who aren’t.