The first piece of AI-generated video I ever made moved me to tears – tears of laughter. Given the chance to fool around with Runway AI’s Gen-3 Alpha, I dropped in an image of an eagle carrying off a wolf. Moments later, the picture sprang into life. The eagle slowly flapped its wings as it glided down a mountainside, dropping the wolf from its talons. Except the bird only had one leg – and its plummeting prey sprouted wings from its tail and morphed into a wolf-headed goose. It was weird and hilarious.
Make no mistake, though – this is the future. Generative AI has given us amusingly surreal images such as the pope in a puffer jacket and a 90s nightclub where everyone is Gordon Ramsay, but the entertainment industry is not laughing. In fact, it’s panicking. A recent statement opposing “the unlicensed use of creative works for training generative AI” has been signed by more than 25,000 writers, actors and musicians, including Julianne Moore, Kazuo Ishiguro and Thom Yorke.
It’s not just performers; AI threatens massive upheaval to people who make our movies, television shows and video games. A report published in January predicted generative AI would disrupt more than 200,000 entertainment industry jobs in the US by 2026, and 75% of respondents believed the technology would lead to job losses in their area. One visual effects (VFX) executive describes the situation to me as “like King Canute trying to turn back the waves. It’s out of the box and it’s never going back in again.”
Things are moving rapidly. This February Open AI, the creators of ChatGPT, unveiled Sora, their astoundingly sophisticated text-to-video tool. In response, actor and film-maker Tyler Perry immediately cancelled an $800m studio expansion he was planning in Atlanta. “I don’t have to put a set on my lot,” he said. “I can sit in an office and do this with a computer, which is shocking to me.” Last month, Runway AI announced it was partnering with Lionsgate Studios, makers of the Hunger Games and John Wick movies, to develop “cutting edge, capital efficient, content creation opportunities”. Sony has also said it will use AI to “produce both films for theatres and television in a more efficient way”. New video-generation rivals are springing up: Meta’s Movie Gen, Luma AI’s Dream Machine, Adobe’s Firefly, China’s Kling AI.
And if any further harbinger were needed, in September James Cameron joined the board of Stability AI, creators of market leader Stable Video Diffusion. Cameron laid out the AI worst-case scenario 40 years ago in The Terminator, but now he’s learned to love it. “I’ve spent my career seeking out emerging technologies that push what’s possible,” he said. “The intersection of generative AI and CGI image creation is the next wave.”
Now that wave is threatening to flood an unprepared industry, washing away jobs and certainties. How do people in the industry feel? To find out, I attended Trojan Horse Was a Unicorn (THU), a digital arts festival near Lisbon in Portugal. Now in its 10th year, THU is a place where young artists entering these industries, some 750 of them, come to meet, get inspired and learn from veterans in their fields: film-makers, animators, VFX wizards, concept artists, games designers. This year, AI is the elephant in the room. Everyone is either talking about it – or avoiding talking about it.
“Art should be created by humans,” says João, a Portuguese games designer in his early 20s, as he doodles in a sketchbook (despite being in digital arts, most people here still draw, paint and sculpt). “Every brushstroke comes from experience, from hardship. AI takes out the curiosity, the learning.”
“It makes things faster, but I think it’s more dangerous than useful,” says Rosa, a Spanish concept artist sitting next to him, as she draws in her own sketchbook. “Bottom line: it’s stealing our work and it’s taking our jobs.”
“I’ve only been in this industry for three years and I’ve already lost two jobs to AI,” agrees another round the table. Before, she was doing illustrations for games concepts, she explains, but thanks to AI, “the CEO decided, ‘I can do this on my own now, we don’t need you.’”
Up-and-coming artists used to start out with work like this, illustrating pitches for new projects to studios. But as another young artist tells me, “Pitch work is dead. They’re not hiring artists to do that any more. They’re going to generative AI.”
The games industry laid off more than 10,000 people in 2023, and this year’s cull is reportedly higher. Companies have already been using AI for “five or six” years, an insider tells me. Activision Blizzard, the Microsoft-owned makers of games such as Call of Duty and Candy Crush, has already laid off 1,900 workers this year. In January, Riot Games, makers of League of Legends, laid off 11% of its global staff.
It’s a similar story in Hollywood, which is increasingly digital in terms of big-budget animations and special effects-driven movies. Now the industry is looking battered by declining cinema attendances and expensive but poorly performing movies – compounded by last year’s actors strike, which was partly over AI. Hence the talk of “efficiency”.
“It can do things that used to take two weeks in less than a day,” says the VFX executive, who did not wish to be named. An ever expanding array of AI-powered tools are now at film-makers’ disposal and they are irresistibly powerful: “We were taking just pencil sketch outlines, and [using AI] you can make them look 3D, so you get rid of the need for a modeller to sculpt something for you. It is frightening.” People like him are in an impossible position, he says: “I’m making my own obsolescence, because if I don’t use it, I’m obsolete; if I do use it, I sort of functionally bring an end to myself as well.”
Some artists at THU are taking on the AI companies directly, like Karla Ortiz, a fine artist who also does concept art for high-end movies. Her work shaped the look of Marvel’s Doctor Strange, for example – the character, the costumes, the hair, the mannerisms, even, arguably, the casting of Benedict Cumberbatch. Going one step further than the statement by Julianne Moore and co, in January 2023, Ortiz co-launched a class action lawsuit against a group of AI companies including Stability AI, claiming they “violate the rights of millions of artists” by training their models on artists’ copyrighted work – without their knowledge or consent.
Ortiz, who is Puerto Rican-born and San Francisco-based, first stumbled across AI-generated art online in early 2022. Having dug into the training datasets, she believes these companies assimilated the work of every artist she could think of, including herself: “I found almost the entirety of my fine artwork in there, and I knew that’s what powered these models.” None of the artists were informed, consulted or compensated. “My paintings aren’t just copyrighted – they’re my life,” says Ortiz. “This feels like identity theft.”
She believes a more ethical form of AI is both possible and necessary, but even she acknowledges that it will be difficult to put this genie back in the bottle. “Once the models are trained, they can never forget. Un-learning is not a thing … A lot of people say, ‘Well, it’s out there, adapt.’ Here’s how I adapt: I’m gonna sue the living pants out of them.”
Stability AI and the other defendants have successfully dismissed aspects of the class action, but legally this is uncharted waters, and the lawsuit is now proceeding to discovery, which means Ortiz’s side will get to see internal communications within these AI companies, with a view to eventually going to trial. A victory could have enormous repercussions. “I think we have enough to totally win,” Ortiz says.
From the AI companies’ point of view, rather than destroying the creative industries, they are saving them. “The cost of making content has skyrocketed,” explains Cristobál Valenzuela, co-founder of Runway AI, speaking over videocall from the US. “People are spending hundreds of millions of dollars to make films, and that’s not sustainable long term, because the market needs more content. So something needs to happen. And for us, this is the moment where technology can come and help drive the cost of production down.”
Born in Chile, 35-year-old Valenzuela studied economics, taught himself software engineering then came to art school in New York, where he and two friends founded Runway in 2018. “We’ve always been fascinated by the intersection of AI and art,” he says. Since 2022 the company has held an annual AI film festival in Los Angeles and New York. Despite my disastrous wolf-eagle experiment, Runway’s AI is already being used in film and music videos, including the Oscar-winning Everything Everywhere All at Once.
What data Runway trains its models on is “confidential”, Valenzuela says, but their deal with Lionsgate avoids at least some of the copyright issues raised by Ortiz, since it is using the studio’s own back catalogue. “If you have content in one particular style, the model can better assess and understand that style, and so help you generate in that particular direction.”
That doesn’t mean you’ll be able to make a movie just by typing in, say, “John Wick 17”, but it will make it easier, he says. “There’s a lot of fires and explosions and CGI in those movies, right? Well, blowing up cars is really expensive.” Even blowing up a CGI car “needs hundreds of hours of time and work. Wouldn’t it be great if we have a system that does it better and faster?” He says AI “will free you from much of the tedious, repetitive day-to-day work. That’s great. We should celebrate that.”
This doesn’t necessarily mean a reduction in personnel, reckons Valenzuela: “I actually think we’ll see the opposite. If the cost of making stuff goes down, then you will start hiring more people, because now you can make more projects.” Some jobs will disappear, he argues, but new ones will be created. “We used to employ people in elevators to press the buttons, or people to throw rocks at your window before alarm clocks were invented.” He thinks “Will AI replace humans?” is the wrong question. “It’s software, it’s a computer, it’s technology, but a pen is technology. And so it would be strange for us to position an argument like, ‘Do you prefer humans or pens?’”
Veterans of the industry have seen this kind of technological upheaval before. When CGI movies such as Toy Story arrived, old-school animators felt they had become obsolete. Many left the profession, but others adapted and learned new skills. As one industry pro at THU points out, CGI actually grew animation: a traditional hand-drawn animated feature required about 120 artists; last year’s Spider-Man: Across the Spiderverse employed about 400. What’s his AI advice to THU’s young artists? “Before we get fucked by it, we have to make it ours.”
Andre Luis, the 43-year-old CEO and co-founder of THU, acknowledges that “the anxiety is here” at this year’s event, but rather than running away from it, he argues, artists should be embracing it. One of the problems now is that the people eagerly adopting AI are executives and managers. “They don’t understand how to use AI to accelerate creativity,” he says, “or to make things better for everyone, so it’s up to us [the artists] to teach them. You need people who actually are creative to use AI.”
Luis likens generative AI to ultra processed food: it cannot create anything new; it can only reconstitute what’s already there, turning it into an inferior product. “And a lot of companies are trying to make fast food,” he says. Many see AI as a way to churn out quick, cheap content, as opposed to higher quality fare that has been created “organically” over time, with loving human input.
In addition to Ortiz’s lawsuit potentially cutting off the supply, the public doesn’t like AI-generated content, and nor do the artists working in these industries – many of whom feel like trained chefs being made to flip burgers. “There are a lot of people quitting these companies trying to do AI fast food,” says Luis. “They are searching their own paths in a different way, and I think they are right.” He sees THU as part of that; beyond the festival, it is also a year-round platform for creatives to connect and collaborate.
The democratising potential of AI could usher in what Luis calls “a new era of indie” in films, games, TV. Just as digital technology put cameras, editing and graphics tools into the hands of many more people. “AI will allow a lot of young kids that never had the budget to implement their ideas, to do incredible stuff,” he says. “Instead of being scared of AI, a lot of young creators are thinking, ‘My dreams are now possible. I don’t need $100m to do this – I can do it with $2m.’ Instead of one organisation with 500 employees, you’ll have 100 organisations with five employees.” As happened with cinema in the 90s, “the major studios will start buying content from the indies,” Luis predicts.
This might not be good news for entertainment corporations, but it could be great news for audiences, creators, and art itself. What’s important, says Luis, is not the bottom line but culture and the impact it has on people, to enrich society. “AI is something that is here,” he tells the young creators at THU, “so you need to adapt. See the opportunities, see the problems, but understand that it can help you do things in a different way. You need to ask yourselves, ‘How can I be part of that?’”
Source link
lol
