As the animation industry faces a contraction, one force is stirring up more fear and anxiety than any other: artificial intelligence. The Annecy Film Festival will be screening four AI-based films after receiving dozens of submissions using the technology.
Concerns that AI will eliminate some animation jobs or replace humans entirely have sparked discussions about how AI will impact artistry as well as workflow. As issues around copyright, creation, and overall usage are ironed out, some see AI as a tool that must be mastered and applied with care. With all the questions it raises, it can be seen as just another technological evolution. Just as a generation of artists had to grapple with Photoshop and its impact on digital images, this generation of artists will have to learn to use AI.
“At the end of the day, a computer is an expensive pencil,” says Cathal Gaffney, managing director of Dublin-based Brown Bag Films (Dylan's Playtime Adventures). “A computer is a creative tool that helps artists realize their vision. It enables artists, producers and directors to create. The people who will lose their jobs to AI are the ones who don't know how to use it to do their jobs better. AI is a productivity tool. It speeds up certain tasks. The industry is struggling right now for a variety of reasons. AI may be the worst thing people are worried about, but I think universities have an obligation to start educating students on the benefits it can bring in terms of speeding up some tasks.”
In March, at Ireland's Animation Dingle comics festival, Gaffney participated in a panel called “The Rise of AI in the Creative Sectors,” which was designed to give students aspiring to work in the animation industry a chance to ask industry leaders about how AI will impact their futures.
“Everything at the festival is programmed for students,” says Maurice Galway, director and co-founder of Animation Dingle, “and we speak to a lot of young people who don’t like the student programming. [AI]But it's not going to change. It's not going to go back, so people are going to have to learn to use it as part of their job.”
AI tools like Stable Diffusion (used for image generation with text prompts) and Midjourney (often used for rapid prototyping) have upended expectations of what AI can do, but they are not as developed as some hope or worry. The market is flooded with other image generation tools, such as DALL-E, ImageFX, and NightCafe. More such AI-based tools are coming to the market. But human animators and creators are still the main source of ideas, and will remain so for some time to come, according to Delphine Doreau, program director of animation at Pulse College in Dublin. But animators will need to know how to use the tools to keep their skillsets up to date.
“AI uses data to make things pop up,” says Doreau. “There's no futuristic vision. There's no research, no future, no creativity. So there are limitations. Artists can't be replaced, because as much as you want to do something innovative, you can't. But this is not so obvious in today's market. So the risk is either stagnation, like in the Middle Ages, when innovation was dangerous, or using AI as a tool to help you work a little faster.”
AI also raises thorny issues around copyright of intellectual property: stories, characters and entire animations often take years to develop, from pre-production through to iterative production processes, where scenes are screened repeatedly, workshopped and rewritten.
When original creators feed materials into an AI and teach it how to create things like backgrounds so that animators don't have to repeat certain tasks during production, it's no different than the old Tom and Jerry cartoons where the backgrounds play over and over when the mouse is trying to escape the cat. But when companies or individuals other than the original creators try to use materials they don't own in conjunction with an AI, the situation is completely different.
Drew Mullin, head of production at Canada's CBC Kids, believes broadcasters are concerned that AI could raise copyright issues with data collected from unlicensed sources. For producers and artists, this represents an opportunity to use tools to speed up production time, but carefully controlling what material is used to train AI for specific tasks will remain a challenge. For now, broadcasters are considering how to respond.
“These types of machine learning tools, you know, have been around for a while. [much] “The teams that we work with have a lot of software that has this built in,” Marin said. “It's amazing, but I think first and foremost we need to protect the artists and filmmakers and producers that we work with. As part of CBC Kids, we are a public broadcaster. We're approaching this very carefully and looking at everything on a case-by-case basis, because this technology isn't going anywhere. [AI] “How it's incorporated and how it's used is very important. I think every broadcaster around the world is now looking at this and considering implementing policies.”
Gaffney believes production companies have an obligation to discuss how they will and won't use AI in their work. He plans to post his company's policy on his website in the coming weeks, which will influence what types of AI skills Brown Bag wants its animators to have and how it will leverage AI technology. The policy will change over time depending on how tools evolve and how artists use AI in their work.
“All animation studios have [AI] “Just as you have an environmental policy, you also need a safety policy,” Gaffney says. “It's important that both clients and staff understand the company's approach. There have been small changes before, but this is a major disruption.”
Gaffney points out that something like Midjourney is a “generative AI trained on the copyrights of everyone from Pixar to Disney to Brown Bag Films.”
“So anyone in the world can say, 'Give me a character design that looks like Doc McSteens or Pixar characters.' The computer is learning the copyrights of a bunch of different people. So essentially, the computer is mimicking other people's copyrights.”
It's an ethical issue. “If you're in the business of creating copyright, you also have to be in the business of respecting other people's copyright,” Gaffney says. “And a lot of the generative AI that was trained on old copyrights is [artists’ and studios’] Copyright is no different than a studio using pirated software. Automation isn't going away anytime soon. How we use automation will change, but we shouldn't disrespect other people's copyrights at our own expense.
“There are boundaries that should never be touched.”
