Issue link: https://digital.copcomm.com/i/1504940
more. Then the director says the problem is the breath, so he goes into the recording booth and he records dog breaths at three in the morning. True story, by the way. If someone is hoping that AI can replace us, they're still going to need us to run the AI. Because there's too much of what we know that I think is unlearnable. I think it's about feeling. KK: In that case, you'd have to request the variations to AI over and over again to try the different approaches. SB: I've heard a story, not sure it's 100% true, but it does make a point. There's a scene in Annie Hall when they're fooling around in bed and she has to stop because an ambulance goes by. In the mix, the sound editor puts a siren in because it's in the script. And the director, Woody Allen, says, "Hey, what's that?" "It's a siren because the script says he stops having sex because of the siren," and Woody Allen says, "Oh no, take that out." So, it's a big thing about what people are used to hearing. It's hard for AI to do "art" [in this context]. I mean, if we imagine asking an AI to take the Mona Lisa and re-work it in the style of Jackson Pollock—is it better? Is it the Mona Lisa anymore? AI is going to be wonderful for certain things and we're going to quickly realize that AI is not wonderful for other things. When we're mixing, we use the [phrase], "Does that sound 'pull you out?'" Does it take you away from being lost in the film? And if that's not your intent—if the intent is to pull you out, like when Ferris Bueller looks at the screen and starts talking to you, that's intended to pull you out, right? But if it's not intended to pull you out, that's what an AI could do—it could make the dog collar too loud. KK: Part of our job is how we present the sound to our clients. Sometimes they don't want to accept certain material because they need to be taught about what they're hearing. They need to be prepared. Further, how would AI deal with notes that aren't actually based on anything audio? We've all gotten notes based on insecurities. They might even be based on "I really want an extra week of mixing, so I'm trying to sabotage this so I can collect some time." And I'm not naming names! SB: [Laughs] Oh, then I'll name names! I sound designed a Paul Schrader film, and he was amazing. But at the end of the mix, he just wanted two more weeks. So, his notes led to two more weeks. And in the end, it made the mix much better—he was correct. Here's the quick analogy I have for this. When I have a client's project, I'm super comfortable saying to them, "Okay, look, we've been over this really, really well. We've reviewed it a few times. We could watch it again and make the dog barks louder, or lower the collar jingle. But it's great. Now you've got a great mix and I feel like the more we do, the more likely we are to undo something good." I mixed and sound designed the second film from an absolutely crazy talented woman. She took 15 years between two films. I think one in 10 million humans could have survived that mix. An AI? No way would the director have been satisfied. The trust and the bond that we build with those directors when they say, "Is that loud enough? Is that appropriate? Is that too loud? Am I making a mistake here?" That's the creation, right? That's us creating art and some directors don't need that, but I think [many] do. The good ones hire people like us so we can give them trusted, valuable advice. KK: The fact is that for more and more people, TV has become something that we listen to while doing something Steve Borne, CEO of HearLabs