- AI tools are considered helpers to human activities such as brainstorming, said executives at South by Southwest.
- Throughout the conference, attendees imagined what chatbots, deep-fakes and content-generating software will mean for creative industries. Actor Tye Sheridan is developing AI to ease motion capture.
- Users are warned to beware of inherent bias in AI data sets
In a theme that will resonate across Hollywood, the majority of users at stock photo site Shutterstock are using generative AI “for inspiration.”
“They’re using it to get rid of that blank page problem,” said Meghan Nally, chief product officer at Shutterstock. “They’re using it as a starting point.”
The idea of AI as “a universal basic intern” (an assistant that can be used in your daily work, whatever your line of work) was the overall theme at SXSW, where Nally was speaking. The partnering with AI tools as co-creator seems to be the overriding opinion of those companies pushing AI tools in the media and entertainment industry.
Shutterstock launched generative AI capabilities earlier this year and now has two million new images being generated every week.
“These are images that didn’t exist before,” Nally said. “These are things that only existed in people’s imagination that then they’ve been able to bring to life in a matter of seconds.”
At a live presentation at SXSW titled “Generative AI: Oh God What Now?“ two technologists, Romain Bogaerts from Real Chemistry and Sameer Grover from AbbVie, pondered how many creativity-driven jobs will get taken over by machines. In a Shark Tank-esque pitch session, entrepreneurs proposed new ways to integrate AI into entertainment, such as by splitting audio stems or visualizing film scripts automatically. A SoundCloud executive told another audience that people who categorically reject AI-generated music sound “a bit like the synthesizer haters” of electronic music’s early days.
“AI could be an amazing tool to help democratize a lot of the aspects in filmmaking,” actor Tye Sheridan told Brian Contreras at the Los Angeles Times. “You don’t need a bunch of people or a bunch of equipment or a bunch of complicated software with expensive licenses; I think that you’re really opening the door to a lot of opportunity for artists.”
Along with VFX artist Nikola Todorovic, Sheridan founded Wonder Dynamics, a Hollywood-based company focused on using AI to make motion capture easier.
In a demo Sheridan and Todorovic showed the LA Times, the software took an early scene from the James Bond movie Spectre — of Daniel Craig walking dramatically along a rooftop in Mexico City — and scrubbed out the actor to replace him with a moving, gesturing CGI character. The benefits, to Sheridan, are straightforward.
“I mean, you don’t have to wear those silly-looking motion capture outfits anymore, do ya?” he said.
BuzzFeed chief executive Jonah Peretti believes this isn’t another bubble like NFTs or stereoscopic filmmaking that is destined to burst. The rise of AI is more akin to mobile phones or social media, he said, “massive trends that changed the economy and society and culture.”
Also at SXSW, Amy Webb, CEO of consulting firm the Future Today Institute, imagined a world in which AI programs are used to mass-produce many different versions of a single TV pilot, either to focus-test them before release or to show different ones to different viewers after.
“I bet sometime in the next handful of years that there becomes this horrible industry practice where you have to have multiple variations before things are greenlit,” Webb said. “And then there’s a predictive algorithm that tries to determine which version has the highest likelihood of grossing the most [money].”
The rise of AI in writing has also raised concerns by unions representing screenwriters, who fear studios might replace experienced TV and film scribes with software. According to the LA Times, the Writers Guild of America will demand studios regulate the use of material produced by artificial intelligence and similar technologies as part of negotiations for a new pay contract this year.
Fred Werner, who works for the UN information and communications agency ITU on a program called AI For Good, was eager to talk about inherent bias in AI data sets.
He said, “When you’re developing these tools, do they work equally well on men and women or on the elderly, on children or on people of different skin colors or people with disabilities or in low resource settings and in the least developed countries? These are not questions that occur naturally to the fast tech startup industry [where] it’s more like build it first and fix it later.
“This is important because you need to create a kind of common framework of understanding if you’re going to connect AI innovators with problem owners.”
A “problem owner”* could be a local mayor, a doctor, an NGO, a teacher, a scriptwriter. The question for AI For Good is how can people engage with AI tools in a way that they know data sets are inclusive.
“You’re mindful of things like gender and disabilities and where people live. And right now there’s a digital divide. As useful as these tools are as feeling like having ten interns in my pocket, that’s just going to increase the digital divide for people who don’t have it.”
Understanding that outputs are only as good as the inputs, Nally explained that Shutterstock is attempting to address that concern. She said the company is applying “logic” between the input and the output that allows them to filter for things like hate speech.
“It ensures that we’re generating results that are diverse. That we’re mitigating bias. It’s not humans that we’re [using] to help get to the place where you’re generating visuals that are representative, that aren’t full of hate,” she said.
The industry is going to have to learn how to apply that filter of machine logic on top of the basic AI models, Nally said.
Nally also addressed the way Shutterstock is handling royalty payments on generative content.
“Whenever a piece of content is generated and downloaded on our platform, we pay a royalty to a contributor fund just like we would direct to contributors. That’s incredibly important to us because our business model does not exist if we’re not inextricably linking supply and demand creator, artist, contributor. So we’ve been really thoughtful to ensure that whether this is 50% or 5% of content budgets in the future, that our contributors will have the ability to continue to make money off of their content in a variety of ways on our platform.”
Shutterstock has more than 600 million assets in its library not all of them get sold. But AI promises to help monetize the long tail.
“We have contributors who are uploading their work and have not ever made a dollar on Shutterstock. Now all of a sudden they have the opportunity to make money because of this technology. It’s [no longer] just how effective you are direct selling to a consumer,” said Nally.
“I think we are the only company in the world that’s created this ecosystem today that truly connects generative outcomes with contributor compensation.”