Does it Matter if the Story that Moves You Was AI-Generated?
Only if you care about genuine, meaningful connections with people and their work.
NY Times journalist Ezra Klein recently asked a guest on his podcast, in reference to online content: “should it matter if it's done by a human or an AI? Or is that some kind of…sentimentality on my part?”1
I want to think about his question in the context of creative content more broadly. AI-generated books, short stories, opinion columns, podcasts, blogs, paintings, poems, TV, film, and just about any other expression of creativity. Does it matter – in some ethically charged sense of the word “matter” – whether those things are AI generated? And if you think it does matter, is that, at bottom, foolish sentimentality?
I think it does matter and I don’t think it’s due to sentimentality, foolish or otherwise. It matters because the value of our engagements with creative works depends on our being engaged with the human intentions, thoughts, and feelings that generated the work. To get to this conclusion, I’m going to start by thinking about our more direct relationships with people.
The Quality of Our Relationships
You may have seen the movie “The Truman Show.” In it, Truman lives a blissful life. He’s happily married, spends quality time with his best friend, and has perfectly amicable neighbors and coworkers. But unbeknownst to him, he’s living in a massive TV studio. Everyone he interacts with – his so-called wife, friend, and neighbors – are actors.
Things begin to unravel when a spotlight comes crashing down from what appears to Truman to be the sky but is in reality a kind of IMAX dome. Eventually Truman ascertains the truth, is rightfully appalled at the deception, and figures out how to get the hell out of there.
Prior to the discovery, Truman was happy. He felt good about his wife and marriage, his friendship, and his life more generally. He felt satisfied, joyful, and connected to his loved ones. If you looked inside his head, you’d find an easygoing, contented man.
But the crucial question is not whether he was happy. The crucial question is whether Truman was living a good life before his discovery. And more specifically, did he have a good relationship with his wife and his best friend?
To these questions, we have to say ‘no’. Despite what he believed and felt, he had a soulless marriage and an empty friendship. His ignorant bliss makes his life all the sadder.
Truman agrees with this assessment. What he discovered is that his life is a bad one. He didn’t find out that he’s been unhappy all along, in the way that someone wakes up and realizes they haven’t been in love with their spouse for years. What he realizes is that his happiness was not well-founded.
The same line of thought applies to his relationships. Truman’s wife treated him nicely on the surface. But underneath we can see that her professions of love, her friendship, her advice were not genuinely expressive of love, friendship, or an interest in his well-being. She was acting. She was faking it. That’s what renders their relationship of little to no (if not negative) value. All this despite Truman’s adoration of her.
The moral of the story is this: you can’t look inside someone’s head to see if their life is going well. You can’t look inside the head of one person in a relationship and assess how well that relationship is going. How well we are doing in life and love depends upon factors outside of our selves and our perceptions.
Our Emotional Engagements with Creative Works
We don’t only have relationships with people. We also have relationships with creative works. And our engagements with them are not simply a matter of what’s going on insider our heads. We are engaging with a work that was created by someone. The art isn’t incidentally associated with the artist, as if it’s a flake of the artist’s skin we’ve come to admire. It’s no accident that the art looks or sounds or feels the way it does. The work – and again, I mean everything from fine art to podcasts to cartoons – expresses or manifests or flows from the intentions, thoughts, and feelings of the creator. And the intentions, thoughts, and feelings of the artist explain or ground, at least in part, the value of the art.
You can see where I’m going with all this. If Truman’s wife’s actions are fake in the sense that they don’t genuinely express the internal things they purport to express – the intentions, thoughts, and feelings of a loving wife – then so too are AI-generated creative works fake in the sense that they don’t genuinely express the internal things they purport to express. AI simply doesn’t have any intentions, thoughts, and feelings to express.
Of course, AI generated work may well bring about wonderful thoughts and feelings inside of us, just as Truman’s wife brought about wonderful feelings inside of him. But this isn’t enough to tell us about the value of the relationship. This is at least one way – I would say the primary way – that it matters whether we’re engaging with human or AI-generated content, even if we can’t feel the difference from the inside.
Two Objections
I’m putting forward a controversial view. I think there is room for reasonable disagreement. I’ll address just two objections here.
The first objection is an attempt at a compromise. “I’ll grant you that AI generated art that we interact with has less value than if it actually expresses someone’s intentions, thoughts, and feelings,” someone might say. “But the feelings it generates are so rich! That AI generated podcast or story or book or image is so powerful! How can you say there’s no value in that experience?”
But this objection concedes my main point: that it matters whether we’re engaging with something created by a person. This objection says it matters but it’s not the only thing that matters and the experience isn’t completely bereft of value if a person isn’t involved. In a way, that’s fine by me. All of that can be true and it can also be true that it matters whether a human created the work we’re engaging with.
The second objection is that my comparison of Truman’s emotional engagement with his wife is a bad one. Such an objector might argue as follows: “Truman was deceived into thinking he was engaging with someone who was being sincere instead of being a paid actor. So yes, if the audience of AI generated creative work thinks it’s done by a human, then the comparison with Truman’s situation is apt. But for all those cases in which it’s clear to the person enjoying the creative work that it’s AI generated, the emotional engagement with that art is on the mattering up and up.”
I think this objection misses its target as well. That’s because it fails to distinguish between an emotional engagement with a fake, on the one hand, and being transparent about its being a fake, on the other. Let me explain.
Truman’s wife’s actions are fake because they don’t express how she really feels. Similarly, creative work that expresses human feelings are fake because they are not expressions of a creator’s feelings. It’s the fakeness of Truman’s relationship with his wife that undermines (or at least severely dampens) the value of that relationship. Similarly, the fakeness of a person’s relationship with AI-generated art undermines (or severely dampens) the value of that engagement. Deception on top of that just makes it worse. To bring the point into relief, think about a counterfeit Picasso. The art dealer can be honest or dishonest about its being a fake, but whether he deceives or not, it’s a fake, and our relationship with the artwork is thereby tainted.
In Conclusion
Not every piece of creative content calls for the kind of engagement I’ve been talking about. Most AI-generated content is merely informational or just for fun. But our relationships with people are like that, too. Those relationships are fine. They can be fun. But that’s all they are. The deeper relationships with people and with art, however, the ones where we engage with the intentions, thoughts, and feelings of people, are the ones that make for valuable, deeper lives.
If you like this article, you’ll probably like my podcast, Ethical Machines.
https://www.nytimes.com/2024/04/05/opinion/ezra-klein-podcast-nilay-patel.html?