Mind and Iron: When a Computer Starts Imitating a Great Artist
Miyazaki madness. Also, Altman vs Bateman. And what do AI experts think will become of our humanity?
Hi and welcome back to another tasty episode of Mind and Iron. I'm Steven Zeitchik, veteran of The Washington Post and Los Angeles Times, senior editor of tech and politics at The Hollywood Reporter and lead olfactorist at this newsy perfumerie.
Every Thursday we come at you with a slew of AI, tech and future news, telling you what you need to know and what you might fear/thrill to. Please consider supporting our independent mission.
It has been a remarkable week in the land of AI content.
Last week OpenAI released a tool that allowed anyone to create AI images in the style of, among others, Hayao Miyazaki and Studio Ghibli. The tool caught on by the millions — and worried a lot of people who think handmade art should stay true to its name.
Just a few days later, Justine Bateman — one of the most outspoken artists on the subject of the perils of AI (besides Miyazaki) — held a film festival in which no AI content was allowed, firing bac on behalf of art untouched by machines. I broke all this down in The Hollywood Reporter, and will share some bits of what we concluded here.
Also this week, how will AI affect our ability to be human? How will it affect our thoughts, emotions, soul, our sense of self? If you’re a longtime reader of this space you' know this is a question we spend a lot of time thinking about. A lot of futurists and global tech experts spend their time thinking about it too. Researchers at Elon University's Imagining the Digital Future Center just asked them what they thought. We'll share their findings, which were pretty consequential.
First, though, the future-world quote of the week:
“Artists are tubes through which the universe — God, magic, whatever you want to call it — comes through to us. But that doesn’t come through with AI. That’s not a tube connected to that source.”
— Actor-filmmaker Justine Bateman on what she thinks machine-generated art will do to creativity
Let's get to the messy business of building the future.
IronSupplement
Everything you do — and don’t — need to know in future-world this week
Altman, Bateman, Miyazaki; What will AI do to our human traits?
1. THE REVOLUTION WILL BE PROMPT-ENGINEERED.
That was the takeaway from a series of extraordinary developments in the world of creativity and AI this week. And while Gil Scott-Heron may be scandalized by what went down, that’s where it went nonetheless.
The first was that Ghibli development. OpenAI added a tool to its 4o model which enables images to be created with eerie resemblance to a particular studio or style. Consumers — either because they like Miyazaki’s dreamy aesthetic or because the chance to take something that’s normally painstakingly handmade and whipping it up with a few keystrokes is too tempting to resist — immediately set out to make everything from family photos to random selfies look like Ponyo, Ashitaka and other iconic Miayazaki characters. Millions did this, prompting Sam Altman to say the demand was “Biblical.”
What to make of all this? As we wrote in our story in THR:
“At the heart of the Ghibli craze is something good — people’s desire to connect to a fiercely loved artist, the images a testament to the depth and scope of Miyazaki’s influence. But at the root is also something very fraught, a willingness to see all art as both interchangeable and purchase-able. The frantic rush to turn our memories into a personalized Miyazaki greeting card is, through one lens, a testament to our love for the artist, but it is even more pointedly a testament to our love for ourselves, and seemingly no copyright concern nor sanctity-of-art instinct can stop us from fulfilling it.
“In a way, GhiblAI is the end point, or at least the latest distant point, of a trend that has been emerging for a while now. If cinema for more than 100 years was about the portal to a faraway world (who represents that better than Miyazaki?) and social media for the last two decades about putting ourselves at the center of the drama, OpenAI’s new tool brings them into perfect convergence. No longer do we have to see Ponyo’s relationship with his savior Sosuke as the stuff of ancient history and magical lands. It’s right here, right now, unfolding at last December’s sixth-floor office Christmas party. The next step in our self-possession culture is not just dropping ourselves into the story but painting it in the hues of the great artists who came before. It’s turning a vast cinematic history into just another wristband we wear hanging out at the digital mall.”
(You can read the full story here.)
Needless to say, Justine Bateman worries greatly about developments like these. Here’s an extraordinary paragraph from my conversation with her, which happened before the Miayazaki craze went down.
“What I believe will happen is AI will subsume the entertainment business because it helps the conveyor belt. They can now customize based on all the years of user history they have on you. For an upcharge, they can put your head on Luke Skywalker’s body for a showing of Star Wars tonight. Or they know you like, whatever, panda documentaries and Hong Kong fight movies, and so they can combine it and make you a movie….This is like Kleenex, they make a movie and throw it out and make another tomorrow.”
She then added that artist/tube quote above, saying that while the real Tom Cruise consciously does crazy stunts as an act of artistry for our benefit, “no one drove a motorcycle off a cliff to make Deepfake Tom Cruise.”
(You can read the full interview here, which I urge you to do.)
So where does this leave us? Well, some of the biggest tech companies will be racing ahead to create this stuff and not stopping much to count the cost. I don’t know if I agree with everything Bateman said — awkward co-existence seems more likely than subsuming — but she sure lays out a compelling case for its general tilt.
And then just after Bateman said it, the Ghibli OpenAI tool came around, demonstrating in real time where all this might go —and how important it is that voices like hers will be on the journey.
2. THE EFFECT THAT MACHINE-MADE ART WILL HAVE ON US AS A SPECIES is significant; if we're not creating or valuing something new that humans did and are instead relying on machines to generate from the old, we could stand to see some big changes to what makes us human.
But what traits specifically will all this machine intelligence impact? What will the entrance of these non-human intelligences do to/for our sense of curiosity? Our creativity? Our identity?
Lee Rainie, who runs the Imagining the Digital Future Center out of North Carolina's Elon University (they do a lot of cool work; check them out), polled some 300 experts this past January about the topic, asking a number of up-down questions about whether AI ten years from now will have made our existence as humans better or worse.
He also asked this open-ended pie-slice: "Over the next decade, what is likely to be the impact of AI advances on the experience of being human? How might the expanding interactions between humans and AI affect what many people view today as ‘core human traits and behaviors?’”
We aim to have more from Rainie next week. But today, a quick summary of his findings.
About half the respondents said that AI will change our humanity for the better AND worse — that is, for every good thing AI does for us as a civilization it will do something bad, and vice versa. A compellingly nuanced answer (and one, ironically, that makes a case for being human; we do love those paradoxes).
But it's the other half that's really telling. Out of that 50 percent, only 16 percent said AI's effects on us would decidedly lean to the better. A full 23 percent, on the other hand, said it would be mostly to the worse (six percent don't expect a change).
Put another way: 79 percent of the people polled said there would be no net positive to our humanity from AI. Which creates an interesting dichotomy: the great majority of people who study what tech companies are doing don't, on balance, think there's anything to be gained by doing it.
So what about specific traits — on which of those can we expect a positive impact and on which a negative? The study queried for 12 of them. Three saw more positive scores than negative. They were:
curiosity and capacity to learn
decision-making and problem-solving
innovative thinking and creativity
The other nine were negative — AI would ding these areas. They were:
Social and emotional intelligence
Capacity and willingness to think deeply about complex concepts
Trust in widely shared values and norms
Confidence in native abilities
Empathy and application of moral judgment
Mental well-being
Sense of agency
Sense of identity and purpose
Metacognition
Two points jump out, one on a negative impact and one on a positive.
The negative: “empathy and application of moral judgment.” There has been a spirited debate among ethicists and psychologists lately on what impact AI will have on our sense of compassion.
On one hand, the chance to interact with all kinds of new intelligences might seem likely to have a beneficial effect on our ability to feel for others. If before someone who lacked certain emotional-intelligence skills couldn't connect with other humans, the chance to interact with a human-seeming program — the chance to adjust a companion to their emotional meter — might enhance their ability to function as an emotional and thus compassionate being in the world.
On the other hand, if we're spending so much time talking to machines — telling it to pick up our dry cleaning, asking it for psychic validation — how much human compassion can we really have?
According to this study, at least, more experts come down on the negative side of that question. Spend more time with machines, our empathy goes down. Given how much time we'll soon be doing that, and how many billions of people across the planet will be adding machines to the mix, this would seem like no small finding. And there might be a particularly large impact in this country. A famous study in 2016 out of the University of Chicago and Indiana University that looked at global rates of empathy found that the U.S. ranked seventh out of 63 countries —very high. One wonders, then, what will become of it amid our nation’s headlong dive into all this machine intelligence. Social media already pretty documentably reduces our empathy. The idea that we'll be in a cocoon of AI only raises the flags further.
The second point is on the positive side, and offers a note of hope. One of the fears about a machine thinking for us is that we'll become less curious (why try to learn when the machines know?); less problem-solving (the AI Agent can handle that); and less creative (see under: GhibliAI). What a surprise, then, to find a plurality of experts convinced these all three of these human traits will go UP in the age of AI.
Look a little closer and you’ll see why — if machines can do all of these things, we'll have to work harder to make ourselves useful. If the AI is solving problems we once tangled with, we'll have both the opportunity and incentive to take on greater conundrums. On the other hand, if a machine can't really be creative, then that will be an area where we might want to distinguish ourselves. A burst of light, it turns out, in an otherwise dim set of predictions.
Finally, I wanted to supply one of the answers to the open-ended questions about AI’s effects that Rainie posed. It comes from Nell Watson, a machine intelligence engineer and president of the European Responsible AI Office and an AI ethics expert with IEEE.
She writes:
"By 2035, the integration of AI into daily life will profoundly reshape human experience through increasingly sophisticated supernormal stimuli – artificial experiences engineered to trigger human psychological responses more intensely than natural ones. And, just as social media algorithms already exploit human attention mechanisms, future AI companions will offer relationships perfectly calibrated to individual psychological needs, potentially overshadowing authentic human connections that require compromise and effort.
“Most concerning is the potential dampening of human drive and ambition. Why strive for difficult achievements when AI can provide simulated success and satisfaction?…The key challenge will be managing the seductive power of AI-driven supernormal stimuli while harnessing their benefits. Without careful development and regulation these artificial experiences could override natural human drives and relationships, fundamentally altering what it means to be human. This trajectory demands proactive governance to ensure AI enhances rather than diminishes human potential.
“These supernormal stimuli will extend beyond social relationships. AI-driven entertainment, virtual worlds and personalized content will provide peak experiences that make unaugmented reality feel dull by comparison. There are many more likely changes that are worrisome:
“Virtual pets and AI human offspring may offer the emotional rewards of caregiving without the challenges of the real versions."
“AI romantic partners will provide idealized relationships that make human partnerships seem unnecessarily difficult.
“The workplace will be transformed as AI systems take over cognitive and creative tasks. This promises efficiency but risks reducing human agency, confidence and capability.
Well, so much for that light.
3. FINALLY, PRESENTING THIS THREAD WITHOUT COMMENT
The Mind and Iron Totally Scientific Apocalypse Score
Every week we bring you the TSAS — the TOTALLY SCIENTIFIC APOCALYPSE SCORE (tm). It’s a barometer of the biggest future-world news of the week, from a sink-to-our-doom -5 or -6 to a life-is-great +5 or +6 the other way. This year started on kind of a good note. But it’s been pretty rough since. This week continues that theme.
HAYAO MIYAZAKI IS THE LATEST ARTIST TO GET AI-IFIED: Where it goes…we think we know. -3.5
A META-POLL OF FUTURISTS SAYS AI WILL HAVE A MORE NEGATIVE EFFECT ON OUR HUMAN TRAITS THAN POSITIVE: -3.0