Transcription
[music] slop. It's what we've decided to call the countless AI generated images and videos that have started to flood the internet. And it's pretty easy to disregard most of the slop we encounter as annoying, cringe, mildly amusing, or fairly inconsequential.
But the same force driving the creation of slop is encroaching on cultural artistic spaces like film and animation. When I encounter AI generated images or videos, I don't know about you, but I get a very strange feeling. I have a kind of gut reaction that something is wrong with these [music] images.
There is part of me that is skeptical of this gut reaction. Am I falling into a trap that so many generations have fallen into before me? Am I, as the proponents of AI might suggest, just reflexively rejecting the next logical step in technological advancement because it represents an uncomfortable change? Am I merely repeating the pattern started by Plato when he critiqued the development of writing by arguing that it would cause humans to cease remembering?
Or does my gut feeling that something is wrong about art generated by AI speak to something deeper [music] and true? Does it reveal something about what art is for us as humans? And that whatever art is, trying to generate it using AI violates that thing in some way.
This is Tilly Norwood, an AI actress created by Particle 6, a UK-based company that builds themselves as Europe's leading up-and-coming AI production studio. Tilly caused an immediate strong backlash from many in Hollywood. Actors frustrated at the idea of being replaced in an already competitive industry, with some seeing the artifice of Tilly as unsettling on a deeper level. Emily Blunt reacted to Tilly by saying, "Please stop. Please stop taking away our human connection."
Even if the technology is not yet developed enough for Tilly to show up in the next season of House of Dragon, it's pretty clear that some in Hollywood are not opposed to the creeping advance of AI. Some talent agencies have reportedly been circling Tilly. Ridley Scott has said he's trying to embrace AI. And Bloomhouse's Jason Bloom argued in reaction to Tilly Norwood that AI is here to stay, whether we like it or not, and that Hollywood shouldn't stick its head in the sand, but should instead embrace AI as long as it can be done ethically and legally. He made sure to clarify. Ashton Kutcher called AI an amazing tool that we should learn to work with.
And this is the most common refrain in support of using AI in art, that it's just another tool, that rejecting AI would be like rejecting digital cameras or computer animation in film. And it's this argument that the creators of Tilly use in their response to the backlash. I see AI not as a replacement for people, but as a tool. She is not a replacement for a human being, but a creative work, a piece of art.
But is Tilly art? Could Tilly really be used to create art? Is AI just another paintbrush we can use to put strokes on the canvas? Or is it something else?
There are many potential reasons to be critical of AI. Environmental concerns, legal and ethical concerns, even mental health concerns. And while I'm worried about many of these, my goal here is not to address the entire complex issue that is AI as a whole. I'm going to focus on one small piece of this puzzle, a relationship between AI and art.
I'm focusing on AI and art because I think this debate forces us to clarify our human relationship to this technology. And it's part of the discussion that is important to me. But we can't understand if AI can be a tool that artists can use, if it's something we should totally reject, or some balance in between, if we don't understand what art is and what its purpose is for humanity and society in the first place.
[music] In Cave of Forgotten Dreams, director Werner Herzog takes us into the Chauvet cave in France [music] to see the oldest paintings ever discovered. J.F. Martell, writing about the film in his book Reclaiming Art in the Age of Artifice, says, "In the images this prehistoric people have bequeathed to us, we get a glimpse of something like a shared humanity [music], but we also gaze into a stranger part of ourselves [music], something reaching to the depths."
Seeing works of art made some 30,000 years ago, there is a striking sense of familiarity. There's something in the activity of creating art that feels [music] instantly recognizably human. An impulse which remains within us despite such a vast passing of time and how much society has changed across that span. [music]
What we find looking at this art is the mystery of who these people painting these images were and why they painted them. [music] It's as if we discovered signs of alien life only to realize the aliens are us. That our own existence is just as mysterious and startling as anything else in the cosmos. Even if we cannot know what these cave paintings mean, there is a profound and moving mystery in the question of their meaning itself.
These images don't just represent the existence of something or someone, [music] but a self-awareness of that existence. The existence of the art in the cave [music] is existence expressing itself for others to see and hear.
What we don't often think about is that despite how much art has changed, despite all the technological advancement and scientific knowledge we've acquired since these cave painters first did their work, we are still caught in this mystery today. Art for us is still mysterious, and true works of art still express [music] the mystery of life itself.
According to Martell, art demands that we feel and think the mystery of our passage through this body on this earth, in this universe, and bears witness to the bafflement that the mere fact of existence elicits in our brains. What is the impulse that connects the intricate carvings in Angkor Wat, these cave paintings, and the Sistine Chapel with the mournful beauty of Debussy's Arabesque Number One, or the hauntingly poetic images crafted by Andrei Tarkovsky in *The Mirror*? [music]
I see in all of them an attempt to express something the artist experienced but struggled to quantify. [music] An attempt to express a mysterious part of life that cannot be put into words. A desire to leave not just a mark on the world, but our mark. [music] A mark that says something about who we are and our experience of life. Art is the thing that allows us to communicate about these mysteries, to share between two humans the unquantifiable aspects of our existence.
OpenAI, the company behind ChatGPT, recently released Sora 2, an AI model that allows you to easily generate AI videos that sometimes look startlingly real. But it is also a platform that allows you to view an infinite stream of these artificially generated videos. Scrolling through Sora induces a kind of dizzying, uncanny feeling in me. There is, of course, something familiar about this. It looks at moments like I might be watching a TikTok or Instagram reel, but there is something alien and strange about it as well. At times, it evokes a chuckle, more at the nihilistic absurdity of the thing itself than any given video or joke created by the platform.
But the overwhelming feeling I get looking at Sora, other platforms that feature streams of AI generated images, Tilly Norwood, or reading a poem generated by an LLM, is a strange sense of emptiness. There is, to me, a perceptible vacancy to the entire thing. While the occasional AI image might initially grab my eye, I'm not drawn in by the same curiosity I feel when a human work of art grabs [music] my eye. I do not want more. I'm simply left asking, "Why does this exist? What is it for?"
You can ask an image model like Midjourney to generate a cave painting in the style of the Chauvet cave, and you'll get something like this. Setting aside the fact that this obviously wasn't skillfully crafted onto the contours of rocks some 30,000 years ago, it would be easy to critique this image by pointing out which details it gets wrong. Tallying up all the ways in which the AI images do not measure up to the original in terms of quality or style. But I think doing so would miss the point, because this image, even if it got the details right, would still be missing something that the original paintings [music] have.
But what exactly is missing? If this AI image isn't art, then what is it? What do we call it? To answer these questions, I'm going to draw on the distinction made by J.F. Martell between art and what he calls artifice.
In our lives, even long before the advent of AI, we have been surrounded by and inundated with designs, photographs, paintings, illustrations, graphics, symbols, videos, animations, and other creative works. Within this huge landscape, we find the stuff we typically call art. But we also find the same skills and techniques that are used to create art used to create a lot of stuff that is something else. Most of us would probably agree, for example, that Van Gogh's *Starry Night* is a work of art, but would probably not think of a stop sign or an event flyer as art in the same sense.
Martell calls this other category of thing artifice. It's important to note that artifice itself is not bad. The distinction here is not a value judgment. And there is obviously artistic craft and skill involved in the design and creation of artifice. Intent, creativity, composition, color theory are all employed to create an effective road sign or event flyer. So this distinction is not about diminishing the talent that is often needed to create good artifice or the necessity of artifice within our society.
Martell says, "Artifice foregoes the revelatory power that is art's prerogative in order to impart information, be it a message, an opinion, a judgment, a physiological stimulus, or a command." The distinction between art and artifice lies in the purpose they serve, both for the creator as a means of expression and to the viewer.
When I look at a work like Edward Hopper's *Nighthawks*, there is no obvious message or statement to be found. Instead, I find a strange sense of [music] mystery, a weird juxtaposition of liminality, loneliness, and beauty that is hard to describe, but which feels familiar to me. [music]
I get the same sense from Hopper's *Morning Sun*. Is this loneliness, despair? [music] Is the warmth of the sun about to awaken her from this melancholy? Or is it somehow the cause of it? As if the warmth of its touch is making her long for something more. Is the emptiness on her face, which is reflected in the sparseness of the composition, connected to the industrial landscape that clouds her view from the window? Is Hopper suggesting that these emotions are the result of the modernist environment that was reaching maturity in the 1950s? I don't know. But he captures here in these paintings a feeling that I recognize. And in recognizing that feeling, I feel connected not just to the painting, but beyond that, to the human expressing feelings through the painting.
Now, let's compare *Morning Sun* to these images created by Karl Lagerfeld for a 2010 campaign [music] for the fashion brand Fendi. There's obviously a necessary level of artistic skill needed to craft and style these images, [music] but unlike Hopper's original work, there is no mystery at their core. >> [music] >> We do not need to wonder what the images are trying to express, what the model is feeling, or what emotion the photographer is trying to convey. We know the purpose of these images is to showcase the clothing, to get us to associate these clothes and the brand with the prestige of Hopper's art. The images are trying to get us to buy [music] something.
While I like these images, and I think Lagerfeld is obviously a talented photographer, Martell would argue, and I'd agree, that they are not art, [music] but works of artifice. Artifice isn't made for the purpose of exploring some deep, mysterious feeling. Artifice might use or evoke feelings, but its ultimate purpose in using those feelings is to convey information or to get the viewer to do something. [music] Whether that's drink a Pepsi or yield to oncoming traffic. And doing this can be important. You don't want a stop sign to express the profound mysteries of the universe. You just want it to clearly and effectively convince the viewer to take an action.
Authentic art, by contrast to artifice, takes us out of the realm of the utilitarian altogether. Instead of trying to get us to do something or convey information. According to Martell, art astonishes and is born of astonishment. Astonishment, he says, is the litmus test of art, the sign by which we know we have been magicked out of the practical and utilitarian enterprises to confront the bottomless dream of life in sensible form.
In this view, art is made and exists simply as an expression of and way to encounter the mysteries of life. And the astonishment that defines art can come in the form of grand expressions of religious wonder or more subdued, subtle examinations of emotion. It can come in something as small as attention to the way light plays across curtains blowing in the wind, or the simple beauty of how objects and color dance together in a room. It can even show up in a self-aware study of the way large expanses of tone and the texture of brush strokes can evoke certain emotional states.
Martell explains that this astonishment comes not from transcending the uncanny strangeness and mystery of our everyday experience, but by awakening us to it. Astonishment is not necessarily a conscious end goal that the artist sets out to symbolize within the work. It is the driving motivation, something that comes through the process of creating art itself, which is revealed both consciously and unconsciously in the thousands of decisions made by the artist in the act of crafting a work.
Each culture or individual has their own taste in what they value, and a stylistic language that they use for expressing this astonishment. What astonishes one person or culture may confuse or bore another. But what all art has in common is this attempt at expressing our own astonishment at some aspect of existence itself. It can be astonishment in the face of profound joy or the depth of agony that can exist in life, and everything in between. And while not every work of art will awaken this astonishment in every viewer, when art is serving its true purpose, astonishment is there. [music]
I don't know if the painters in the Chauvet cave were expressing a sense of wonder, fear, [music] desire, worship, or celebration when they were depicting these animals that would have been their source of life. Perhaps what led them to create these images was the way all of those feelings about the subject coexisted inside of them. We can only speculate. What we do know is that there was something astonishing in these painters' experience that drove them to take time away from all the activities in their lives that served a more obvious survival function to engage in the act of expression. And whether or not we get our interpretation of their images right, [music] much of the value we get out of the images themselves comes from trying to connect with what they were expressing.
So what happens when we have an AI generated image of a cave painting, or any other kind of image that looks like art? It might be visually similar, just like the Fendi campaign bears a resemblance to the Hopper paintings, but as an image object, I would argue the AI generated images do not carry the same expression of life that true works of art do. AI, unlike the cave painters and other artists, cannot express astonishment at some aspect of life through art because it is not itself experiencing that life. It is not feeling astonishment which it can then translate into the artistic process. Anything it makes can only be an imitation of the end result of humans engaging with the artistic process. An imitation made with no understanding of what it is imitating or why.
You can ask an AI to generate an image of what loneliness and alienation looked like in the 1950s. And you might even recognize and resonate in some way with how AI depicts those feelings. But these images are not an actual artifact that reflects the real human emotions that someone felt in 1950. While Edward Hopper's paintings are. And I think the extent to which we might connect with the emotion of an AI image is simply the extent to which we connect with its ability to imitate how human artists have already translated experiences into art. It can generate infinite variations of these imitations in every conceivable style, but it cannot add to the corpus of art something new born out of a direct contact with experience. It does not have its own experience of loneliness to then express.
And what draws me to a Hopper painting is [music] not the content of the images themselves, not truly the end result, but what is behind them. A painting is not just graphic wallpaper meant to decorate a room or a museum. I don't look at his work to see a city street or a woman sitting in a bed. What draws me to Hopper is his perspective [music] on these scenes. The way he captures certain feelings through how he constructs [music] and paints the everyday. He saw the world in a certain way and expresses that perspective through his work. [music] And his work connects me with another human's experience, while the artifice only connects me with the arbitrary elements of style that Hopper may have used to communicate his experience. [music]
I believe much of what draws us to art is the artist's perspective. Literally getting to see [music] how their way of seeing the world gets encoded into their work. AI fundamentally does not [music] have its own perspective on the world. There is no way of seeing that it can then encode in a work. At least in its current iteration, it can only produce [music] imitation of existing perspectives. That might work for the creation of artifice, but I think it leaves out one of the most significant reasons we are drawn to art.
The heart of this issue is not just found by evaluating the end results. It's also found by looking at how the works are made. The artifice of AI generated images or text is baked into the very way it is created. Beyond conveying information and commands, Martell argues that artifice works in service of creating consensus. Consensus is the statistical world of useful knowledge, generalization, habit, custom, and ideology. Works of artifice reinforce consensus by representing reality as though everything had already been mapped out.
The machine learning algorithms that generate text and images literally work by using statistical and mathematical analysis to try to create predictive data. An AI generated work is not created using the artistic process. It's created by visualizing consensus found in data and information. AI uses this consensus to reverse engineer an image in order to meet the desire or expectations that we give it through the prompt. The AI is a visual expression of the very consensus that Martell argues is antithetical to art.
So while we might use it to generate something new in the sense that it might be a particular arrangement of pixels or words that has never existed before, it seems to me that AI cannot be a means by which true astonishment and mystery are captured and transmitted. Because true astonishment and mystery are not pieces of data or information. They are something much stranger, felt, and experienced. True art is the part of life that cannot be turned into data or information. That's the whole reason why we encode these things into poems, paintings, or songs in the first place instead of just describing them using logic and reason.
But why should we care so much about the distinction between art and artifice? If we lose sight of the dividing line, we risk becoming victims to two unfortunate misunderstandings. First, we might treat works of art as artifice and lose out on allowing ourselves to experience the astonishment and connection that authentic art has to offer. Or we treat works of artifice as works of art and leave ourselves vulnerable and exposed to the ways artifice is trying to influence our thinking and behavior.
Our society is currently in the grips of both of these misunderstandings, beyond the ways we're manipulated by advertising. We fall prey to all manner of propaganda. And we simultaneously devalue the power of art by frequently asserting that everything is artifice, that there is no room for something greater that can truly move us profoundly.
Martell argues that in our modern world, we are often closed off to allowing the forces of art to profoundly affect us as a defense mechanism against the manipulative power of all the artifice that surrounds us. But this very guardedness that we use to protect our mental health in such an artifice-saturated world has turned into an apathy towards art, something we need to thrive and connect as humans. We put on ironic detachment and see being unaffected by authentic expression as cool. While we view those who are moved by something as weak, sentimental, and romantic become insults.
This jadedness has affected even the artists themselves, who often only engage with anything sincere through a layer of detachment or irony, not wanting to be seen as cliché or sentimental. The cost of all of this is high. Cynical artifice posing as art, concerned only with its own commercial viability or ability to persuade or influence people, and numbed-out audiences unable to truly connect with the rare artists and works that do manage to get through the landscape of cynicism and exist as works of authentic art.
I focused here mostly on images generated entirely by AI. But what if AI was just one small tool used in the process of making a larger work of art? This is, at least currently, the question that faces most filmmakers. I don't think any small piece of artifice within the larger work necessarily corrupts the entire thing. I don't personally think that just because a small amount of AI was used on part of Adrien Brody's performance in *The Brutalist*, that we should treat the entire film as if it has no artistic merit and that it should be called artifice.
There seems to me an obvious gap between that and Tilly Norwood becoming a reality. And perhaps, as Jason Bloom suggests, if we can iron out the ethical and legal issues with AI tools, we might be able to use them as a tool in our toolbox, in the same way that CGI and digital cameras are used. But I think we should be very hesitant to do so, because what makes art a work of art is not just the end result, but the process we use to get there.
True expression in art emerges from every tiny choice made in the process of bringing that art into reality as a representation of our experience and perspective on life. The artistic process is something both conscious and unconscious that reveals itself through each brush stroke. Through accidents and how we respond to them, through collaborative decisions made by humans, through the way the environment we make the art within subtly influences each choice in the process. Each decision or accident we make in the process of creating art is an opportunity for meaning, perspective, and experience to sneak into the work, even when we're not consciously planning for that to happen.
Any aspect of the process that we use AI for then necessarily removes you from the process. But you and your perspective is the very source of what makes something art. The prompt is not a tool you use to express yourself through. It's a way of describing an outcome you want to achieve and having something else try to achieve that outcome for you.
In the documentary series *The Directors*, Leonardo DiCaprio shares a story from the making of *The Aviator*. In one scene, both he and Scorsese knew that something wasn't working about the performance. And Scorsese, after many takes, told him that, but did not want to tell him what to do to make it work. Because Scorsese knew it would be more valuable and more authentic in the performance if DiCaprio found what worked on his own. Scorsese did this because, as an artist, he understands that the process matters. That even when working with a human, sometimes giving them an explicit outcome to work towards can be antithetical to the artistic process.
If the essence of a work of art is often found in the very process of attempting to get to an outcome, AI disconnects us from that process. So, if we want the machine to make our art for us, it strikes me that we may not understand the purpose of the art we're making, that the true value of that art is not a perfect outcome that we imagine AI will help us achieve, but the process we take to get to an outcome. And when we use AI, we risk undermining the very process that allows deeper meaning to emerge through our work. A typewriter might have made writing words more efficient, but it did not write the words for you. AI risks doing too much. And in that process, artistry loses out.
If you want to make true art rather than just artifice, I think you have to engage with the artistic process. Not so that you end up with a perfect finished work at the end that makes the experience you're trying to express comprehensible, but in order to have a work that brings the audience along on your attempt to express the incomprehensible. This is why looking more closely at a work of art reveals deeper meaning, while looking closer at a work of artifice often reveals it to be less meaningful than it originally seemed.
But I must admit, while I have not yet encountered an individual AI image that feels like a work of art to me, sometimes I do feel a kind of dreadful astonishment at what these images represent. If we peer into this endless hall of mirrors constructed by some of the largest corporations on Earth, which attempt to reflect everything we've ever created back at us, and ask why, what do we see?
When I look at it this way, I see a work that astonishes me with its absurdity, surrealism, emptiness, and glut. [music] An attempt to create a work that encompasses everything humanity has created while simultaneously trying to remove [music] as much of ourselves from the creative process as we possibly can. I see humanity trying to abandon its own involvement in the act of creation, leaving itself with only its desires, its prompts, waiting to be filled with increasing accuracy, precision, and speed. And maybe this attempt reveals a hidden desire for self-negation. That [music] perhaps our desire to remove ourselves from the creative process itself is born out of a belief that we are fundamentally flawed. [music] That we must overcome those flaws by creating a machine that transcends our imperfect abilities. One that can infinitely meet our needs with the touch of a button without ourselves or another human being needing to be involved. That can put an end to the unbearable mystery of it all.
But what if we have it backwards? What if our involvement in the process, as messy as that may be, is the very thing we desire and are trying to satisfy with this endless [music] cascade of novelty, information, and increasing productivity? [music] What if we are seeking a machine that can perfectly fill our prompts when what we truly desire is for ourselves to be prompted by an artistic vision? To care enough about something, an experience or perspective on the world, that we want to do the creative labor ourselves, to falter through the decisions and learning process, finding the expression of that astonishment in each brush stroke.
What if what we want [music] is an artistic project where no shortcut is fulfilling? Because the project is not a means to an end, but fulfilling as a process in and of itself, as a part of life, which is not a problem to be solved, but a mystery to be astonished by.
Imagine with me that perhaps 30,000 years ago, the Chauvet cave painters finished their work, put down their torches and charcoal, gathered around the fire together afterwards, and one of them said, "Wouldn't it be great if we could create something that would do the painting for us so that we wouldn't have to do it?" It sounds absurd to us when we hear them say it, but in many ways, that's the world we're living in today.
In the interview where Jason Bloom said, "The industry needs to embrace AI," he also said, "The consumer does not care if what they're looking at is AI." And at first, I thought, he's obviously wrong. Some people, like myself, do care about whether or not what they're looking at is AI. But on closer examination, I think he's right. A consumer does not care about whether or not what they're looking at is AI. A consumer is just looking for whatever might be available to fill the craving they have in that moment. Whatever might draw them in, appease them, and amuse them will do the job. And there is certainly money to be made in giving consumers what they want.
But we have to ask ourselves, are we just here to appease consumers and to consume, or do we want something more? Appeasing consumers is not the purpose of a work of art. If you are creating to appease or experiencing something for the sake of being appeased, you are within the realm of artifice. Art is not made for consumption but for the sake of expression and connecting with yourself, others, and existence through that expression.
While there is a place in this world for the creation and consumption of artifice, and that artifice can bring us a lot of joy and pleasure, if we lose sight of the special role that art can play in our lives, both as artists and audiences, I think we lose sight of a valuable part of being human that has been with us since the Chauvet cave. We cannot escape our desire for the connection provided by authentic art by getting better, faster, and more efficient at making art of this.
Getting a clean shave is one of those things that always, to me, feels like a little bit of a risky endeavor. I used to be like, "How bad is my face going to be mangled this time?" But these days, when I do want a clean shave, I don't worry about it because of this razor created by Henson Shaving, my sponsor for this video. It leaves my skin feeling better than any other razor does. This is thanks to the engineering experience and the precision that Henson put into creating this new kind of razor design. I can tell you that this razor will leave your skin feeling better than other razors. But Henson is putting work into actually backing up that claim with research. They partnered with a medical imaging startup, and early research is showing that this razor leaves you with less razor burn than other popular brands. Henson used their experience designing aerospace parts for like the Mars rover and the International Space Station to design a razor that holds the blade with such incredible precision. And that's what gives them an edge, or we could say the least edge, of any razor on the market. The other thing I love about this is it's an incredibly high-quality aluminum razor. You buy it once. The replacement blades you can get anywhere for like three to five bucks for a year's supply, and everything from the blades to the [music] packaging is recyclable. So, you're not just constantly tossing plastic razors into a landfill. I'm happy to recommend this razor to anyone who shaves. And you can get yours when you go to hensenshaving.com/thomas. If you use the code ThomasFlight at checkout, you'll get 100 red replacement [music] blades for free when you purchase the Hensen razor. Again, that's hensenshaving.com/thomas Flight. Use code ThomasFlight to get those free blades at checkout. Thanks again to Henson for making this video and mustache possible.