0:00
/
Transcript

Hollywood Reporter: Joseph Gordon-Levitt on AI, Filmmaking, and the Need for Guardrails

At Sundance, earlier this year, The Hollywood Reporter hosted this conversation on AI, innovation, and independent filmmaking.

Transcript from the video:

Stacey: Welcome. I’m going to call you Joe. You said I could call you Joe earlier. So Joe, it will be. You are very unique to me in a lot of ways. You’ve been very active in this space since the mid-2000s. Twenty-plus years. You started HitRecord, which I remember as the advent of an incredible online community. Which of course became a really wonderful start-up. I’m wondering, along the way, when did you start to have concerns? Was there an inflection point for you where you started to worry about the advent of this technology on the work of artists, including yourself, including acting?

Joseph: Yes, funny. You bring up HitRecord, so why don’t I start there. Let’s go back to 2010, right?

Stacey: Sure.

Joseph: It’s funny, because now the tech industry has become like just a bunch of Lex Luthors. But back in 2010, we all thought the tech industry were heroes. We thought: these are the smartest, best, and brightest minds in the world, who are going to fix this messed up world with their incredible technology. That’s what it really felt like.

Stacey: And social media was a big part of that.

Joseph: Social media was on the rise. There was the Arab Spring. There was Occupy Wall Street. Social media played such an important part of those movements. And I was working on this thing called HitRecord, which you just mentioned, which by the way launched here at Sundance in 2010. That was our first moment as a production company. I’m feeling kind of…Dan used the word “tender” earlier. That really struck me about being here at Sundance.

Stacey: I saw Gregg Araki’s new film the other night and I felt very tender for you “Mysterious Skin” was a seminal independent film.

Joseph: Thank you.

Stacey: You’ve been around, Joe. You’ve been around.

Joseph: Me and Noah were in a Sundance movie together back that same year, actually. So with HitRecord, the idea was: can we use social media to do something that was never possible before artistically? Can people from all over the world work on a collaborative art project together? And we were able to do that. It was so much fun. We made all kinds of things. Films, music, books, records, TV shows. It was really wonderful. And I was as optimistic as I think a person could be about what social media was.

And I remember you asked this question: where did it sort of start to turn for me? It was a book. A very particular book I read, by Jaron Lanier, who is one of my favorite thinkers. And I actually read his book because his second book is called “Who Owns the Future.” And in the second printing of that book, there was a new introduction he wrote in the context of the Snowden disclosures. And at that moment, I was busy playing Edward Snowden in a movie. So I read this book.

And Jaron predicted all of this. It’s crazy. Go back and read Jaron Lanier’s essays from like 2001, and it’s like he’s talking about today. It’s really incredible. And he talked about how this business model that social media companies were using, where they don’t charge customers, but instead charge advertisers, give the product for free, maximize the engagement of the users, and manipulate them on behalf of the advertisers. He really broke it down in an incredibly prescient way of how that was going to ultimately break the world. And we’re seeing it.

And then, of course, I think the floodgates sort of opened in 2016. There was Cambridge Analytica. Donald Trump won the election and everyone was like, how did that happen? And it surprised everybody who was so doe-eyed, me included, about technology. We realized, oh, it’s not just that Twitter will help Barack Obama. It’ll also help Donald Trump.

Stacey: We forgot about that part.

Joseph: Yeah, and when you read what Jaron has to say, he actually talked about movements like the Arab Spring and Occupy Wall Street. And he said, those played really well on social media, didn’t they? It felt really good to read through those tweets. What happened? What came of all those posts about the Arab Spring? What came of all those posts about Occupy Wall Street? They weren’t actually productive, because they took place on platforms that weren’t designed for human connection and collaboration and productivity. They were taking place on platforms that were just designed to hook you, keep you, and serve you ads.

So now, we’re facing a new technological revolution with this thing that gets called AI. And what’s funny is a lot of these AI companies will say, oh, this is totally different. Don’t put us in the same bucket as this social media technology, which has now gotten a little bit of a bad name. AI is just here to help.

First of all, a lot of them are the same companies. The same companies that are building these models ran social media. And even the new companies, they’re running the same business models, the same exact business models. Hook you, keep you, serve you ads. And they’re running the same exact algorithms that drive that business model, the algorithms that are so incredible at digesting billions and billions of data points to derive the exact thing that will hold your attention so it can serve you more ads. And we’ve seen the damaging side effects from that with social media, and we’re about to see those damaging side effects get amplified so hard with these AI platforms.

And look, I share the views of what Dan, Noah, and Jenna and Matt from Autodesk were saying. I actually think this technology could be incredible. But the path of least resistance won’t take us there. The default path gets dark. And so now I think this is the moment where we all have to say: we can’t just let it go the way it’ll go if we let them just follow profit incentives. There have to be other guardrails in place.

Stacey: So some of the most vulnerable folks and collateral damage that we see every day are kids. And you’re a passionate parent of three. How’s that going, by the way?

Joseph: Oh, it’s the very, very best thing in my life.

Stacey: Oh, I imagine you’re a wonderful parent. And an entertaining parent nonetheless. You have a great Substack, and you do something called Joe’s Journal. Really great videos.

Joseph: Thanks.

Stacey: And in one of them, you’re talking about screen time, which I know is a constant issue for parents, and also for adults, by the way, but managing kids’ screen time. And you said something really interesting, which was: screen time isn’t the problem as much as the predatory addictive algorithms, which you just intimated.

Can you put into context exactly how these algorithms are impacting little children? As young as we see the babies in the restaurants, and I feel for the parents, they just want to eat, they just want to put a bite of food in their mouth and chew. But then I also feel sick to my stomach, thinking, that poor baby. And it’s no one’s fault. But I’m wondering: what is the impact that you can see now, even in your own kids? And how can we try to cut this off?

Joseph: Whew. There’s a lot there. I mean, if you’ve ever seen those photos of little kids smoking cigarettes back in the 1910s or 1920s, it was common as hell back then. Little kids smoking cigarettes. We didn’t know. People didn’t know. And I think probably today it’s still not quite common knowledge how addictive and damaging a lot of these products are, especially for developing minds. There’s more and more science about this. Read Jonathan Haidt. Jonathan Haidt wrote an incredible book, any parent out there should read it, called “The Anxious Generation.” It’s about a lot of the science around the impact these algorithms have on young developing minds. He also just put out a brand new book for kids. I got it for my kids. It’s sort of like half graphic novel, half illustrated book. It’s really great, illustrative book. So any parents out there, check out “The Anxious Generation.”

But look, one thing, if there’s one single thing I’ve learned about being a parent, it’s that every family is different, and don’t give advice to parents about how to parent. Because we’re all different and no judgment. For me, it’s not that we keep our kids away from technology. We don’t. One of my kids is really into making digital music. It’s amazing. It’s like one of the happiest things in my life ever is checking out the stuff that my kid makes. The tools that he has today to make music or make videos or make animations compared to what I had when I was 10. It’s so fun, it’s so incredible. And I have no problem letting him sit there and work on his music project for as long as he wants to. But that is a very different thing than subjecting him to these algorithms, which are completely designed to be addictive.

You can read “Hooked” by Nir Eyal. It details the social psychology, the science of hooking the human brain, exploiting just the way that our brains work in order to hook you and keep you. And we all like to think, well, it doesn’t work on me. But if it didn’t work on all of us, these companies wouldn’t be trillion-dollar companies. It works statistically. It works in the margins. You don’t necessarily notice it. And it’s the same techniques they use to build slot machines that they use to build these algorithms. Whether that’s an algorithm driving a recommendation engine recommending the next video on YouTube or YouTube Kids, or it’s the algorithm determining the next word in a sentence generated by a chatbot that’s going to keep you, or the next pixel in an AI-generated video. All the same thing.

And it is addictive. And yeah, you don’t give your kids addictive stuff. You wouldn’t bring your kid into a Vegas casino and say, here, have fun on these slot machines. You wouldn’t give your kid cigarettes to smoke or alcohol to drink. Why not? Because it’s addictive.

Stacey: So I think it’s hard because we use the products too.And then we’re modeling the behavior. And you even say this in your Substack, you say you’re not perfect, of course. You’re a busy person. So that’s what’s hard. You can’t say “don’t do drugs” because you yourself have to use these devices. So it’s about mitigating the usage and teaching them about it.

Joseph: I think it is about that, and it is about parents taking personal responsibility. But I would say that in addition to personal responsibility, it’s too much responsibility to put on any one person. There is a collective problem. And any parent has experienced this one. You can try to say to your kid, no, you can’t have this thing, you can’t use this service, and they’re like, yeah, but every other kid in my class does. So you have a coordination problem there, and that’s what laws are for. We should, as parents, hold firm boundaries with our kids. But also I think part of being a responsible parent is to get together to utilize our democracy and say: we collectively, as parents, want this to stop.

Stacey: Absolutely. And my final question for you is a two-part question. For you, what does an ethical future look like for Hollywood, since we are all creatives in this room? And who or what is giving you hope? Because hope is in short supply these days, it seems. But it’s hard to not think of Robert Redford, who brought us together here more than 40 years ago. Without him, we wouldn’t be sitting here. I think he would want us to stay hopeful. So what is driving hope for you, and what does an ethical future look like?

Joseph: Well, I’ll say, look, I’m a member of the Creators Coalition on AI as well, that Dan and Janet were talking about. And I think part of the work of that coalition is to assemble a structure. There are so many things that should define what ethical use of AI is. I’ll talk about one thing. And this again comes back to Jaron Lanier, who’s written really intelligently about this stuff.

These AI products, they’re called artificial intelligence. There’s a great thinker named Divya Siddharth who says, let’s not call it that. Let’s call it collective intelligence. Because really, they don’t artificially generate anything. What generative AI really does is it takes every movie ever made, every book ever written, every article on the web, every song ever put out, and it statistically calculates. It breaks all that humanity into data, into what they call tokens, and finds the statistical probabilities between those tokens so that it can generate these incredible outputs. But it doesn’t work at all without all that human input.

So right now, the tech companies are taking the position that they should be able to just take all that input without permission, without compensation. I think if you take that down the line, that’s no good for anybody. We’re here talking about independent filmmaking. And I completely get why an independent filmmaker would want to use these tools. And when I say independent filmmaker, I mean whatever. Like my ten-year-old and his buddy, or two ten-year-olds in Nigeria or Mumbai or wherever else, who don’t have access to the kinds of film industry that we have access to, who want to use these tools to make beautiful things. And make things in a way that we “hot shots” can make them because we have the budgets. They should be able to use tools like that.

The question I have is just: who makes the money when the tool gets used? That’s all. Use the tools, but where’s the money flowing? Because right now the only place the money is flowing is a hundred percent of the value of these tools going into like four or five companies. And where does that lead? And why should it be that way, when these tools wouldn’t be valuable at all without all of those human inputs?

To answer your question, we need to build a system for ongoing compensation. Not a one-time buyout, but an ongoing system where when a model generates an output, it can calculate what inputs were used, which inputs were used more than others, so it can rank the importance of those inputs and pay people accordingly. We already do it on YouTube, for example. They split their ad revenue in half and pay all their creators. They right now haven’t yet built the technology necessary to track the inputs of these AI models to the outputs. But that technology needs to be built, and then policies need to be put into place. It’s going to be a lot of work. It’s going to be really, really complicated. But that’s the kind of upgrading of systems that I think of when Dan talks about upgrading systems.

And what gives me hope? I’ll say two things. One is that we’re having this conversation. Back when we didn’t see the downsides of social media coming, it was years and years and years before we started having these conversations. We’re a lot earlier on the uptake this time, and that I think is reason to be hopeful. And the other thing I would just say is, I have a ten-year-old, an eight-year-old, and a three-year-old, and they’re so smart. They’re going to figure this stuff out better than any of us.

Stacey: Thank you so much, Joseph. Thank you everyone for coming. So happy to have you. Have a great night. Thank you so much.

Discussion about this video

User's avatar

Ready for more?