Issue link: https://digital.copcomm.com/i/1504731
M OT I O N P I CTU R E S O U N D E D I TO R S 19 DEAR DIALOGUE DETECTIVE occasional tool." Gabe K.: "There's a lot of people making some wild predictions based on very little real-world use case data. I personally think it's too early to tell how it will affect an entire industry from a few flashy demos and a lot of hype from people who coincidentally stand to become obscenely rich from more investment in it. It's like trying to extrapolate how 3D animation would affect the industry as it is today from the Money for Nothing video." Randy Thom (supervising sound editor, Lucasfilm): "AI will dominate almost all of this kind of work eventually. Right now, the quality of AI isn't great, but it will improve enormously over the next few years. The bigger budget/higher profile projects will be the longest hold-outs, of course. My guess is that most of the low- and mid-budget projects will be AI five years from now, and nearly everything will be AI 20 years from now, possibly sooner. People will still be needed to direct and edit the AI, but human staff levels will be drastically reduced." Steve C.: "You probably already know google has a tool for dubbing and ADR. I think it's game over for a lot of voice actors. These tools can make an actor say whatever you want in their own voice and adjust lip flap to match. I would imagine there are similar tools for SFX that will make things much faster and cheaper." https://slator.com/google- teases-universal-translator.../ Martin C.: "I don't doubt voice work is under massive transformation. But even that takes time and patience to work with the world's worst actor. On the positive side, AI won't want to be pampered and ego dealt with, etc.,—and it's always available/sober/ willing, etc., etc. For SFX and Foley, for sure no technical reason, why not? I do wonder how quickly the finances will stack up. Everyone needs voice stuff, not just pro broadcast, so a lot has been sunk into that (think banks, google translate, etc. ) But Foley? Maybe the niche will protect to a degree, and some directors can't be bothered directing a human." Joshua R.: "By feeding a large language model large amounts of video content, along with the corresponding FX, MX, Foley, and DX stems, the language model would easily be able to synthesize new stems for new video just as it does with images now. It will only be a few years before FX, Foley, and music stems will be able to be created and revised within seconds. It may be harder to do this with dialogue but also still possible. In 20 years, our society will likely be completely different. Not only sound for movies, everything." The future of AI is affecting almost every aspect of our lives. From fake news and cloned voices of famous people who are "saying things" they didn't actually say to new content that will be hard to distinguish from reality. Even my niece, Rachel Bieber MLIS, who is a digital assets manager at Lindeman & Associates, has this to say about how to vet artwork for legal usage: "There's a lot going on with AI and the legality of marketability and the agency I work for has shifted at least one full-time person to the pursuit of trying to make sure we're on the cutting edge of what is a significant game-changer. As with any new technology, different companies are approaching AI from some wild perspectives. Here's what I've observed thus far: Adobe appears to be approaching integration of AI into their products in a more add-on approach, using AI as a tool rather than as a separate entity. We're not entirely sure yet, but the marketing looks like the licensing would be similar to that of licensing any other type of Adobe tool like brushes. Brushes are licensed for use, not for output. I would say that Adobe already uses what could be considered AI features when it comes to masking and manipulating images; I am certainly a novice when it comes to how to use Adobe Photoshop to create an image as I'm more of a forensic investigator to try and figure out what- all was used to build a final image. Shutterstock is attempting to create "ethical AI," claiming that their AI images are only trained from assets they already have in their collection, and from artists that agree their images can be pulled from, for a modest kickback. Users don't know what that kickback entails. AI-created images within the Shutterstock site are licensable from Shutterstock, but do not carry the same indemnity against lawsuits because they have no way to guarantee something their AI has created isn't too close to something that belongs to someone else. The agency isn't sure about using either of these, as they are SO new. And as far as I can tell, the third of the big three stock sites (Getty Images) isn't yet touching AI for output—only for image searching using images. I've heard there have been legal rulings that as long as AI-created images are created using programs whose terms allow the users to do