This two-part feature explores the opportunities and concerns about the rise of artificial intelligence in the film and TV industry. Part one, below, examines how the technology is currently being used. Read part two to hear from labour experts on the potential consequences of adopting it too quickly.

Industry stakeholders and experts agree that there’s no question that the fast-faced rise of artificial intelligence (AI) will impact the future of film and television production. It’s just a question as to what extent, and who pays the price or reaps the rewards.

Creatives and broadcasters have already begun to experiment with the technology to see how it can be used — both as a cost-saving tool and a way for small shops to do things that couldn’t otherwise be done.

Chris Knight (pictured right), president and CEO of Ottawa-based culinary-focused producer and distributor Gusto Worldwide Media, is among the early adopters of the emerging tech. He claims that Gusto’s The Wizard of Sauce is the first TV show to use AI to generate on-screen characters.

“AI gives us a way to take characters in new directions and amplify the viewer experience,” Knight tells Playback.

The series is hosted by Kyle Crawford, who is joined by AI-generated characters Professor Saussenheimer, Dr. Sauss and Sgt. Sauce-alot, to deep dive into sauce-making techniques.

To achieve the AI characters, Crawford donned three different costumes and stood in front of a green screen to read the scripted lines of each character, none of which look or sound like him. “We then used AI technology to change his personality and his voice,” says Knight. “The only way you would know it was him is because we give him conspicuous credit at the end of the show — and we paid him for those days.”

Knight says Gusto turned to a large suite of tools that are now easily accessible to creators. To turn ideas into images, for example, they turned to Stable Diffusion XL (a latent text-to-image tool capable of generating photo-realistic images based on a text inputs); Midjourney, which generates images from natural language prompts; or DALL·E mini, software that does AI model generation.

Photoshop AI also offered a suite of photo editing tools powered by Adobe Firefly generative AI – including Generative Fill, which allows creators to add to, remove or replace images in Photoshop with simple text prompts.

Knight says the team also took advantage of platforms such as real-time voice changing software Voice.ai, as well as D-ID, which allows creators to make and interact with talking avatars using generative AI via D-ID’s API or Creative Reality studio.

All of it added up to a vision of the future, he suggests. The old-school way of making content is dead,” he claims. “The world is changing. I’m not saying that the world doesn’t need writers. We employ three full-time writers. I have 45 full-time staff. We create all of our content ourselves. But the characters that we generated for The Wizard of Sauce prior to the advent of AI would have been strictly in the domain of Pixar and Disney and other large animation companies. That’s no longer true.”

Knight says Gusto plans to incorporate AI into two other shows – The Wizard of Whatever, a sequel to The Wizard of Sauce (also hosted by Crawford), and Cirque de Sautée, featuring two clowns preparing classic French and Italian cuisine — with more to follow.

“As we get better at asking the AI system the right questions and it gets more sophisticated in answering them, it’s inevitable that we will continue to push boundaries with AI,” says Knight. “AI is a revolutionary change in the way we structure and tell stories.”

While some have turned to the emerging technology to create projects that they would not have been able to before, others are turning to AI as a way of better engaging audiences.

In the U.S., for example, PBS Kids has plans to build AI into episodes of its upcoming series Lyla in the Loop (pictured left), allowing for AI-assisted conversations between Lyla and kids watching at home. The goal will be for kids to be able to ask the main character questions and have AI offer appropriate responses. Pennsylvania-based Mighty Picnic and Hamilton, Ont.-based Pipeline Studios will produce both the 2D-animated series and its AI-enabled episodes.

Ordered this past January and set to launch in February 2024, the 40 x 30-minute series for children aged four to eight centres on a young girl who looks for creative ways to craft new things and learn along the way.

But “no AI utilized [in the series] is generative,” explains Sara DeWitt, SVP and GM at PBS Kids. Instead, she says, “characters in the episodes will respond to users using AI-assisted intent detection. Humans are a critical part of the development of these interactive episodes, as series writers are responsible for creating all pre-programmed character responses.”

At certain times during the interactive episodes, she says, the characters will talk to the camera and give the viewer a prompt question, which are all pre-scripted and recorded like any other part of the action. AI technology then uses intent detection to listen for the answer. Based on that, the story continues with one of several pre-written responses.

“The goal is to support and engage kids in critical thinking alongside the characters,” she explains.

Before PBS Kids launches the episodes, the broadcaster said it plans to investigate their potential efficacy for helping kids learn and stay engaged with the content – which is an ongoing focus for PBS.

“The project and continuing research has been underway since 2019, thanks to National Science Foundation grants, in partnership with the University of California, Irvine and the University of Michigan, and began with PBS Kids series Elinor Wonders Why (2020),” DeWitt explains. “It is critical to help us continue to deepen our understanding of how children can learn from media in new ways.”

So, whether to take on more ambitious projects or to create engagement that wasn’t possible before, AI tools are already a go-to resource for producers. The real question comes down to where the boundaries get drawn.

North of the border, Canadian pubcaster CBC is trying to get ahead of the issue by establishing its guidelines early. Though the outlet is clear that it will take advantage of the benefits of AI, it says it has no plans to use AI for production.

“No CBC journalism will be published or broadcast without direct human involvement and oversight,” stated Brodie Fenlon, editor in chief and executive director of programs and standards for CBC News, in a June 12 blog post.

The CBC was a founding member of Project Origin, an international initiative that aims to make the provenance of original material clear. In early June, the pubcaster also publicly released its principals for AI use, including an assertion that it would not use or present AI-generated content to audiences without full disclosure.

But the CBC is embracing AI in two other ways, says Allison Sugar, the network’s senior director of platform products and digital strategy. “We use it for machine learning and content recommendation systems, helping our online audience find the most relevant content for them,” she says. “We also use it for operational efficiency, such as creating an automated transcript for an audio podcast.”

AI is also being used to tag online stories with metadata, a task normally performed but still reviewed by a reporter, and CBC is looking to employ the technology to convert text-to-speech for the visually impaired and closed-captioning for the hard of hearing.

“AI is increasingly a part of every tool and service on the market and CBC uses a broad range of machine learning methodologies and techniques depending on the use case,” explains CBC executive director of digital strategy, Richard Kanee. “CBC is constantly evaluating these tools against different use cases, as the market is rapidly evolving. As with everything we do, we assess these tools against security, privacy and protection of
intellectual property criteria.”

“For us, AI is a capability tool to solve a problem,” sums Sugar.

This story originally appeared in Playback‘s Fall 2023 issue

With files from Ryan Tuchow and Brendan Christie