Earlier this year, outside the MIT Museum, a man shook my hand after seeing a reading of my play Machine Learning and told me that what really scared him about artificial intelligence is that it would “destroy our concept of reality,” because of its ability to replicate faces or voices of people and make them seem to have done things they never actually did. What will happen to our reliance on physical evidence to establish truth?
I remembered this reflection a few days ago when my dad (who was here with my mom on one of their annual trips to the city) and I rewatched the 2002 movie Simone: “I am the death of real,” says the titular character, a completely digital actress (her name is a composite of the words “Simulation One”) that has given the career of director Viktor Taransky (Al Pacino) an unexpected second wind. The movie, which was written and directed Andrew Niccol (who also wrote The Truman Show and pulled double duty in Gattaca—another parable ahead of its time) had come to my mind when explaining to my dad why the members of SAG-AFTRA were worried about AI. I was shocked by how much I could recall of it 20 years later (I guess I always loved machine stories) and by how current it feels now. Its plot—that the artificial actress successfully climbs to the top of the movie industry—feels ripped from headlines that were decades in the future when the movie was made, and while the story focuses on Viktor’s efforts to keep Simone’s real identity hidden so that people will not reject their new idol (a concern that now feels kinda quaint), the movie’s conclusion that Simone would herald “the death of real” rings very true.
The movie, like all good genre stories, uses its high concept to amplify earthly concerns; Niccol’s target is Hollywood and its obsession with a fresh face—because what is fresher than an ageless simulation? The film is a farce, unconcernedly skipping past plot holes because its critique is not diminished by them; at one point, the mere suggestion that Simone is at a film premiere creates a stampede of movie executives stepping over each other to catch a glimpse of their darling, whom they’ve supposedly signed (and paid) without proof of existence. In the movie, they believe Simone to be human—but in our reality, it feels like such belief is not necessary for Hollywood suits to behave like animals: SAG-AFTRA reports a horrendous proposal that studios brought to the table that would allow them to scan extras’ faces and use them in perpetuity for any other projects without further compensation. (The AMPTP denied this: they say they suggested using the extras’ likeness however they wanted but only for the same movie that the extra had originally shot… without paying them further, which is still very bad.)
So much has been written about current Hollywood woes (most efficiently, in this New York Times piece on the subject), that I’ll just skip to the point. The fear that is currently gripping us is one that we’ve anticipated for a while, but only now feel is actually possible: that machines will replace us because they can do what we do as well as (or better than) we can—but they do it for free. This latter part of the fear doesn’t figure as much in Simone (though the studios do see savings in their budgets when Viktor informs them Simone does not need hair, makeup, wardrobe, or a stunt double), but the digital Simone’s rise to fame is indeed predicated on her unattainable beauty and talent. She is not sentient, per se; Viktor controls her every move (he even has personal conversations with her in which he crafts both sides—he’s the one who utters the “death of real” line first), and her value to him is precisely that she has no personality with which to interfere with his work, unlike the actress played by Winona Ryder, whom Viktor fires at the beginning of the movie for complaining too much. But her lack of sentience is not a lack of intelligence; we see her learn, as when she apes Audrey Hepburn’s “How do I look?” line from Breakfast at Tiffany’s, or when Viktor infuses her with the combined powers of Whitney Houston and Madonna for her breakout music album. Twenty years ago, Niccol already understood the irrationality in our fear, the thing Hollywood might never get: that machines cannot “do what we do” because they cannot create, only repeat (and, it’s worth pointing out, without paying the people who unwittingly trained them, which is at the core of Sarah Silverman’s suit of Open AI and Meta)
“BUT,” you say, ever the Tracy Flick, “didn’t Audrey, Whitney, and Madonna also have their own inspirations, the people they aped?” Sure! No one creates in a vacuum, we all repurpose to whatever degree. But the singular way in which we mix and match, the things we bring into (and pull out of) our references, makes our creations unique. If you, like me, believe in a God who alone can create out of nothing, it makes sense: our creativity is a divine spark that, while it doesn’t allow us to summon forth something completely new, it does allow us to breathe life into a mix of pre-existing elements to make them our own. If you, unlike me, do not believe in God, I still hope you believe in the individuality of each person, the combination of their genes and life experience resulting in something that cannot be copied, only imitated. My own writing is often in dialogue with what inspires me; Machine Learning nods to The Terminator, I, Robot, and countless other AI stories, and yet I can confidently say it’s not the same as any of them. It bears a unique signature that only I can imprint on it.
Machines, of course, have no unique signature. Like Simone, they can only scour the data that we feed them, looking for hidden patterns. When they find one, it can feel like magic—particularly when that finding reflects us back to ourselves, like the writer who saw all his ticks appear in the writing of his AI replicant (on, in a more mundane example, when Gmail guesses what we’re about to type). But that magic comes from the machine’s ability to process information in ways we cannot due to our limited hardware; it is not the mark of personality. A generative AI can only calculate the probability, based on all the examples it has examined, of what word or sentence would be appropriate to say/write considering the parameters it has been given, and then make a choice based on that probability (for example, that the most likely answer to “How are you?” is “I’m fine, and you?”). It doesn’t know what it’s saying. The only real personality involved is ours—to believe otherwise is to make AI into a fetish.
“Whoa, I didn’t know this was an OnlyFans,” you gasp, and I tell you to get your mind out of the gutter. I mean “fetish” in its original sense, from the Portuguese feitiço, meaning “spell.” It’s the term European colonizers used to describe the hand-crafted idols made by West African peoples; the Portuguese could not understand (their own Catholic trinkets notwithstanding) how the Africans could attribute supernatural powers to something they had made themselves, since logic tells us nothing and no one can produce something greater than it/themselves. I am not a religious or anthropological expert that can tell you how the people of West Africa would have explained this conundrum, but I wish I knew—because cut to six centuries later, and some people are doing the exact same thing with AI. It’s the height of human folly, to believe that we are capable of creating a more perfect version of ourselves. And it’s exactly this fetish that puts us all in danger: when we believe that machines are better than us, we put them in charge of all sorts of things they should not be in charge of, like setting bail or assigning credit scores. And when they inevitably replicate our own mistakes and biases (because all they can learn from is the decisions we made before them), we’ll ignore it and say “The machine said so and the machine is perfect, so we must abide by its judgment”—especially since machines are not really capable of explaining their reasoning like a human could, so they can’t really tell us why they did what they did. (For more on this particular issue, I recommend Cathy O’Neil’s cleared-eye Weapons of Math Destruction.)
“Actually,” you argue (my, are you chatty today!), “I don’t think machines are superior to humans, so it’s not a fetish. I’m just saying that most movies already feel like they were written by an algorithm, so why not just hire one and call it a day?” Dad? Is that you? While he was here, my dad and I faced off about this issue many times—he thinks that humans are basically predictive algorithms anyway and, even if they weren’t, it doesn’t take a genius to write a Hollywood movie.
To the first point, I say: NO, humans are not predictive algorithms, even though this is a very popular argument in the AI world; amongst others, Sam Altman, the CEO of OpenAI (which developed ChatGPT), seems to hold it (read this wonderful article about it, in which linguist Emily M. Bender shouts for the people in the back: “I’m not going to converse with people who won’t posit my humanity as an axiom in the conversation.”) It is, I think, absolute nonsense to say that our intelligence doesn’t amount to much more than a parroting algorithm. To compare the irreplicable complexity of human consciousness to a series of equations is to abscond our responsibility over this planet; if we are nothing but parrots after all, let some other species, I don’t know, fix climate change? Those cow farts definitely played a part, so, maybe the cows should do it? I jest, of course, but much like Bender, I have very little patience for the “humans are not that special, we might as well be computers” argument. Humans suck ass, don’t get me wrong, but we suck ass at the top of the food chain, and we cannot pretend we don’t. There is no intelligence on Earth that comes close to ours, and I can’t believe we have to sit down and debate this.
(Chester looks at me with inquisitive eyes—has the time come for the cat singularity?)
To the second point: yes, absolutely, most movies and TV shows being produced right now already give off a ChatGPT vibe. We’re not talking high art or deep insights about humanity when we talk about 99% of the content being pumped out right now (and I am convinced there’s an algorithm that writes girlboss songs for Selling Sunset, no matter what its makers say). So why not make it official? For one, because someone needs to pay me for my writing, damn it! But that makes me sound like a Luddite who won’t embrace inevitable technological progress, so I’ll move to the second, more important (from Hollywood’s POV) point: because it won’t make money in the long run. AI might write for free, but, as I hope I’ve established by now, it can only create imitations of what already exists. Which, granted, shouldn’t be much of a problem in an industry overrun with sequels, prequels, and spinoffs—BUT, but, if you read that Times article I linked earlier, you’ll find this curious section:
This was the year when moviegoing was finally supposed to bounce back from the pandemic, which closed many theaters for months on end. At last, cinemas would reclaim a position of cultural urgency.
But ticket sales in the United States and Canada for the year to date (about $4.9 billion) are down 21 percent from the same period in 2019, according to Comscore, which compiles box office data. Blips of hope, including strong sales for “Spider-Man: Across the Spider-Verse,” have been blotted out by disappointing results for expensive films like “Indiana Jones and the Dial of Destiny,” “Elemental,” “The Flash,” “Shazam! Fury of the Gods” and, to a lesser extent, “The Little Mermaid” and “Fast X.”
Since there are no longer any pandemic restrictions in place for moviegoing, what could account for this drop? My theory: the COVID lockdowns disrupted the inertia that led us to do things just because we had always done them. Looking at the movies mentioned in that quote, it’s not surprising to me that Spider-Man is the only one that did well—it’s the only one I considered watching (though I’ve yet to), because Into The Spider-Verse was a gorgeous movie, a breath of fresh air in a very stale genre. I can’t speak to the quality of the other ones; it never crossed my mind to go see them—I feel like I have already just by reading their titles (in some cases, like The Little Mermaid, that’s probably very close to the truth, it being a live-action remake of a film I know very well). I wouldn’t be too shocked if that’s the same thought other post-pandemic moviegoers had: “Why would I waste my money on this thing that’ll probably not be good? I’ll just wait and watch it on streaming.”
Yes, I hate everything—but eventually so does everyone else, when everything is crappy imitations of good, past things (it took most audiences an extra season to catch up to my feelings about the declining quality of Game of Thrones, but eventually I was vindicated). It’s a cycle we’ve seen many times: something good spawns oodles of ugly larvae. The Golden Era of television experienced by our generation started when a network known mostly for showing boxing matches decided to invest in a weird idea about a mobster who went to therapy; The Sopranos not only made HBO the king of cable, it forever changed the paradigm of TV writing (which up to that point had insisted on characters frozen in never-changing, countless-season-spanning situations). It legitimized TV as a medium and paved the way for some of the best art of this century—but also for thousands of terrible, terrible shows dressed in faux prestige clothing. That’s just the nature of the beast. The Dark Knight legitimized superheroes by being a good movie a decade after Joel Schumacher killed the genre, and look what happened. Eventually, the market saturates with whatever the current fad is (in my lifetime, we’ve had disaster movies, sword-and-sandal epics, medieval trilogies, and YA stories with horny undertones, to name a few), and something new takes its place. This business cannot survive without original ideas that connect with audiences—even if as soon as that happens, the Hollywood golem swallows that idea up and shits out countless terrible imitations. People eat that shit for a while, because they love the original, stan the creators, want to feel the nostalgia, whatever. But they eventually realize it’s shit and stop eating. My certified hater status might make me a Cassandra, turning people off with my negativity, but the most important thing here is that Cassandra was right, and so am I: AI cannot keep the industry alive for long. All it takes is one fad cycle and it’s done.
The “death of real” is a legitimate threat to social order, taking away from us an agreed-upon concept of truth from which to communicate—but when it comes to Hollywood, real was never the issue. Real died when Georges Méliès’s camera jammed while filming a Parisian street in the early 1900s and he discovered, in the editing room, that the sudden cut in the film made it look like cars and people had disappeared mid-movement—and revealing that cinema could be appealing not just by its ability to capture reality, but by its potential to create magic. As far as storytelling is concerned, I’m not worried about the death of real, I’m worried about the death of imagination. It was imagination that gave us the dinosaurs in my fave Jurassic Park, a landmark in special effects that to this day remains unmatched, even by its sequels and especially by any other films (Asher Elbein wrote a great article about this). AI is not gonna give us the next Jurassic Park. It might give us the next Jurassic World, and who wants that?
In Simone, Viktor is at first thrilled by the success he finds with his digital actress, but eventually, he’s envious of how much she overshadows him (she even ties with herself for the Best Actress Oscar!) While casting for his next film, he auditions Wynona Ryder’s character again, who comes back humbled, rid of her diva affectations. Her reading impresses him, and he says she’s so good, she could be the lead instead of Simone. Is it because she kisses him as part of her scene, potentially signaling she’s willing to sleep with him for the part, and he wants to score? Is it because he’s tired of being known only as the guy who directs Simone’s movies? Or is it because, with all her imperfections, the human actress can move him in a way that the machine (even though it possesses all the talent of every movie star that ever existed) cannot?
My bet is the latter. I hope, for art’s sake, that our entertainment overlords come to the same conclusion.