This video of filmmaker Ben Affleck discussing generative AI in a nuanced way has been going viral recently, partly because he has a sharp understanding of the technology -- but mostly, I suspect, because he's one of the few well-known, artistically acclaimed Hollywood figures who isn't actively in Freaking the Fuck Out mode about the coming of AI. (And living in Los Angeles, I can say FTFO seems to be the general tenor among most people in the industry.)
However, I think he's being both too bearish and too bullish about generative AI. While he's definitely right that generative AI will help lower the barriers for low budget indie filmmakers -- it's already doing that -- some of the other applications he mentions are highly questionable. To take two:
"We'll use AI to create personalized versions of our favorite TV shows/movies"
Having written about user-generated content across many platforms for over 15 years, including machinima, I can definitely say this: There's only niche interest in unofficial remixes of well-known IP -- even or especially among hardcore fans. Even the best fan-made movies based on, say, Star Wars will usually generate just a few million views at most (but usually far less).
That's even true of user-generated platforms. To take a local favorite, there's been hundreds of truly impressive Second Life-based machinima videos, but very, very few have attracted over 50,000 views (i.e. 10% of the total SL user base right now).
Why? I doubt it's because of low quality. The best like this Star Wars fan movie are quite well-made. I'd argue that it's the very fact that fan-made projects are not made by the official creators make them seem decidedly not "real" and therefore not worth their time. (Which is a keen irony for the talented fans who made them!)
Or to put it another way: A good test of a true fan is if they know the key people who created their favorite IP.
Try it! Ask hardcore fans of, say, the TV show The Sopranos, the movie Oppenheimer, or the video game Skyrim who lead created the IP. (Or for that matter, Good Will Hunting.) I guarantee they'll cite the writer(s)/director by name, or at least know the key creative studio behind the project. With such close affinity of the creator to the work, the idea that we'll all want remixed AI versions of our favorite movies/TV shows/games seems highly unlikely (beyond a brief fad).
Affeck's other claim that I doubt, though I'm a bit less certain of my skepticism:
"AI will replace thousands of special effect jobs"
I'm sure we'll see some replacement for lower budget projects, but I'd argue the very fact that any special effects scene that's ever existed has already been gobbled up by AI models will create more demand for special effects artists who can create new, untrained effects. We're already at a point where platforms like Sora and Midjourney generate images that seem so familiar, regurgitated, and un-special, we yearn for newness. I also suspect we'll see increasing demand for practical effects, in the same way that the action scenes in Fury Road (mostly shot in-camera with real stunt people) seemed so fresh, exciting, and new.
That point I'm less certain about, so I should check it with some FX artists I know. Many of whom, ironically enough, are busy protesting the rise of generative AI.
Please support posts like these by buying Making a Metaverse That Matters and joining my Patreon!
Comments