Purple Magazine
— The Brain Issue #33 S/S 2020

make love, not code



Artificial intelligence is not a technophile invention poised to spin out of control and destroy the planet, but it could turn the best things about being human into the worst kind of shitty clickbait.

Hugo Caselles-Dupré, Pierre Fautrel, and Gauthier Vernier make something called algorithmic art and, in the process, they also make headline news. Creating a sample set consisting of thousands of portraits and an algorithm to sort them, then assigning aesthetic value to each of them, the collaborators “caused” (I guess that’s how you’d have to put it) the creation of a work entitled Portrait of Edmond de Belamy. The portrait was touted as the world’s first example of AI art. Christie’s sold it for more than $430,000.

The work wasn’t really made by AI. It was generated by a type of algorithm, and yes, there’s a difference, but it’s not worth getting into. That said, I do think it’s worth considering whether AI can generate something akin to the passion that makes an artist unique. That for some reason made me think of Camille Claudel’s relationship with Auguste Rodin, which made me wonder whether AI can machine-think love or something like it into existence. Given the way the human mind and psyche, heart, nerves, sense of smell and touch, and time work, I doubt it. The senses are unknowable — pure subjectivity.

But even if machine learning will never replace our most poignant emotions and creations, it will surely do some of the more tedious things that need to be done in the process of making art and/or love. Killing tedium is good. The path to love can be pretty damned tedious. But process is a lot when it comes to making art or love.

That said, I’m moving too fast.

The first time I researched AI, my filter bubble burped up an international supervillain — the former secretary of state to not one, but two empirically awful US presidents. I consider Henry Kissinger to be a dangerous man. If Henry Kissinger liked AI, AI is dangerous. QED.

But Kissinger didn’t like AI. Personally, I would like AI to suck as much as those billionaires who buy elected officials, dine on endangered animals, and do business with Russian oligarchs who like to give herpes to B-list models. AI can think through things fast — it famously beat the best Go players in the world; Go being famously complex. So it could be good to do complex things accurately, especially when you lean toward the messy complexities of writing or making music or art.

I asked a friend — not a computer scientist but not unknowing — to explain AI to me.

“It’s when one computer, or a network of them, arrives at a decision based on available data,” he said. “It’s all about outcomes and end points. The process leading to something is largely irrelevant.”

This is what Kissinger said, writing about AI and Go: “For our purposes as humans, the games are not only about winning; they’re about thinking.” At the end of the day, whether we’re talking about love or art, it’s all about what happens while you’re trying to get to what happens.

I just read a story about a woman who called the police for help while she was being physically abused by her husband. She told the 911 operator she wanted to order a pizza, and the operator understood what she was doing and sent help. They connected. This is a form of love.

Many candidates for public office practice something called virtue signaling, where they support a worthy or urgent cause to show how awesome they are, but it’s all bullshit. Voters respond — my candidate cares/understands/whatever. That’s a cheaper sort of love. But again, it’s about connecting on a level that is not super-guessable as far as computers go. That said, machine learning can come up with this cheaper kind of love. So, I guess some kind of love or art — whether or not you want to look at a mash-up of a thousand portraits or you’re particularly moved by a politician’s public gestures of altruism — can be produced by AI.

I’m not a computer scientist, so maybe it is irresponsible of me to be writing an essay about AI based on random exposure to it and a willingness to share, but here we are. For me, it alternately figures as a vague sort of threat to humanity and something that could make life easier by streamlining air traffic control, making sex robots better in bed, directing driverless cars, and working out the details of similar tedium that might otherwise go wrong in deadly ways. But when it comes to human experience, which is the bedrock of so much of our creative output, I don’t see it making any huge contributions.

I have a friend who confides to her digital assistant about stuff she would never tell her lover, and AI could for sure use that information, along with the universe of facts known via her online purchases and social media habits, to identify a potential lover she might enjoy more. There’s enough information to create a brute-force attack on our love receptors. AI could probably gather sufficient data about 911 calls to know that a woman asking for a pizza was probably talking in code and needed help.

I am in love and it was hard-won. I have been through different darknesses to arrive at the place where love was possible. Love is famously complex. Neural networks — the AI-generated human brain-like algorithms that conduct brute-force end-point solutions, like improving air traffic control in ways beyond human reckoning — don’t traffic in the unknowns or the variability of the human experience that shift so uniquely in the mind of the lover or the artist.

Maybe it would work backward from the outcome — like starting with the resolution of the conflict in a romantic comedy where all the scenes that came before suddenly make sense and seem like stepping-stones to the inevitable coupling that happens in the movie. If so, my brain can only muster one wordless response: slack-jaw staring into the red barn where we should all go hang ourselves because we now inhabit a world where math molests magic, and there’s no point in going on. Users could pay to have AI find the one they want and make that one fall in love with them, in other words, an arranged marriage.

My grandmother made a beautiful watercolor of an artist at his easel painting my grandparents as they exchanged vows, my grandfather peering back at the painter. The caption read: “Die Liebe kann ich leider nicht malen” (Unfortunately, I can’t paint love.)

Machines can compose essays like this one. But when you Botox the furrowed brow of misunderstood genius, writing is a stupid skill — formulaic and flexible. It’s not hard to imagine someone asking a machine to identify the most accessible angle on artificial intelligence. It could be tasked with finding a controversial application of AI — something likely to go viral. It could game virality using different hot-button topics like gender, race, abortion, immigration, economic inequality. It could study things considered beautiful, touching, horrifying, disgusting, immoral, etc, and deploy them like weapons on the reader’s soul.

In other words, AI could identify true mediocrities of human experience and execute them beautifully. Similarly, philosophy can probably be machine-made — at least certain kinds can. That said, AI-generated Wittgenstein will seem off in a meaningless way. Likewise, a Roland Barthes knock-off won’t reveal the bourgeois ideals found in a bottle of booze, a blouse, or a skirt as keenly as do the now time-frayed discussions of those things in Mythologies.

And it goes without saying that machines can now produce new science without the help of scientists. Human-free factories have been able to make fully automated products for a while. “Human-free” has now extended to more abstract areas like consumer data weaponized to regulate and propagate human likes, wants, and desires in the digital landscape of our new machine-inflected society.

As for artists or lovers being replaceable, I don’t think so. What makes art and love matter is more visceral. It has an is/is-not quality. And of course, love is not so simple and neither is art. It can benefit from an unexpressed conflict from childhood. There are psychological disorders that masquerade as love, and some forms of love that seem like psychosis. A machine can’t approximate these things, because these things exist on a different plane — the process is the point. Love and art are often an unfolding of the unknown or an ongoing celebratory, delicious, conflict-ridden, fermentation of it.

It’s late fall in the mid-1980s, and I am at home with my mother. We live in a house in the country that is set back in the woods about a quarter mile. You can see neighboring houses through the exposed bones of the trees. There’s the smell of damp earth, brown and russet leaves everywhere, and there is one white chicken just off the driveway in some undergrowth. This chicken is a long way from home. It belongs to a French family that lives down the hill.

My mother — a woman I couldn’t stand because I was a teenager and she was my mom, but whom I adored — decides we should catch this hen and return her. We laugh so much. Her sense of fun (and sometimes her way of annoying me) is in some way or another discernable in every woman I’ve ever loved.

It’s a few years later, 1985, and a girl shows up at the first day of school wearing a white cotton dress, a white turban woven through a shock of silky chestnut hair and bright red lipstick. She’d been an exchange student in Yugoslavia. She is exotic where I grew up. She has the same body as the Venus de Milo, but with arms. I yearn for her every minute of the day. After many invented accidental hellos that involved athleticism and planning and a minor car accident, she not only comes to know that I exist, but that I love her. Eventually, she finds this charming and returns my attention. We become lovers — my first time around. She insists we keep the affair secret. People sort of know. It still has to be a secret. We end on a sour note. This story haunts my intimacy for years.

There are so many stories. They make us who we are and inform the way we bond with others. The two stories I just told are now part of my digital footprint because this story will go online, and as such can be used by an AI program designed to find me true love, but any such program would be working with dead matter. The stories taken out of the context of my mind in the world it’s in, as it exists — something like animal being — is not important. I am not math. I am. And who I am, and who any of us are, is impossible to quantify or qualify, much less duplicate or even approximate emotionally.

AI scares me, but not for the reasons that gave rise to Hollywood movies. It’s scary because I think it will demystify love and pain and everything else that make us human, and push us, goad us, make certain people feel exactly right. While the low hum of deep psychology may always be a part of the experience of love for the more self-aware among us, that’s quite different from a machine crunching the facts of our lives, our likes and dislikes, the underpinnings of our emotional life as they are discernible from online behavior, and telling us who to love. Worse, that same process could be used to design the perfect love. It could be deployed with digital assets to lead us to what it deems true love, and keep us there with messages that reaffirm AI’s solution to the problem of who we should love.

Missing is the sacrifice of the ideal that happens as we settle into allegiance with the other, and everything else that makes love so beautiful and tragic, but human. AI will make love no longer love. It will become Disneyfied. Or maybe it will be customizable. Perhaps there can be opt-ins — bugs as features — and lovers can choose the sort of love they want: e.g. conflicted, angry, passionate, short and hard, long and weird, etc.

The terrible man Henry Kissinger wrote: “Before AI began to play Go, the game had varied, layered purposes: a player sought not only to win, but also to learn new strategies potentially applicable to other of life’s dimensions. For its part, by contrast, AI knows only one purpose: to win.”

If you look at Portrait of Edmond de Belamy and find it interesting, you should consider checking out Philip Beesley’s living sculptures, which implement AI but do it in a way that feels integral to something essentially human. I’m fine calling both instances art, but neither interests me as much as the accidents of a Polaroid capturing rotting flowers that my girlfriend accents with beads, because I love her.


[Table of contents]

The Brain Issue #33 S/S 2020

Table of contents

Subscribe to our newsletter