WEBVTT - The Story: Everyone in Hollywood Already Uses AI 0:00:14.000 --> 0:00:16.840 Welcome to tex stuff. This is a story. I'm mos 0:00:16.920 --> 0:00:19.120 Voloscen here with Cara Price. 0:00:19.560 --> 0:00:19.680 Hi. 0:00:19.840 --> 0:00:22.640 Ah, so I would like to show you something and 0:00:22.720 --> 0:00:25.560 just get your reaction, and I promise you it relates 0:00:25.560 --> 0:00:26.400 to today's story. 0:00:27.440 --> 0:00:32.519 Here we are planet Earth. It's impossible not to appreciate 0:00:32.560 --> 0:00:36.680 the sheer grandeur of our world, and yet there remains 0:00:36.760 --> 0:00:41.800 one forest unexplored by humans, a forest filled with life, 0:00:42.640 --> 0:00:46.199 with creatures living in the burrows of the firms, the 0:00:46.240 --> 0:00:50.320 branches of the trees, by the flowing streams, and the 0:00:50.440 --> 0:00:56.920 mossy box. I'm David Attenborough's neighbor Dennis, and welcome to 0:00:57.240 --> 0:01:00.560 a forest filled with little critters. 0:01:01.680 --> 0:01:06.160 So this is a sort of Pixar style animated film 0:01:06.560 --> 0:01:13.160 with a parody of David attenborough narration. The publication date 0:01:13.240 --> 0:01:17.520 is April tenth, twenty twenty three. So I am a 0:01:17.560 --> 0:01:20.200 little curious why you want me to watch this film, 0:01:20.200 --> 0:01:22.560 which is titled Critters with a Z. 0:01:22.840 --> 0:01:26.560 Critters what you just saw is the beginning of a 0:01:26.600 --> 0:01:29.959 five minute animated movie that was created using open AI's 0:01:30.120 --> 0:01:33.440 text to image software back in twenty twenty three. Critters 0:01:33.480 --> 0:01:36.759 with a Z was created using doll e. We're obviously 0:01:36.840 --> 0:01:39.759 now in twenty twenty five, and open Ai, in conjunction 0:01:39.920 --> 0:01:43.240 with a creative agency and film production company, are turning 0:01:43.280 --> 0:01:47.080 critters into a feature length film. Not only that it 0:01:47.160 --> 0:01:49.720 is set to premiere at can in twenty twenty six, 0:01:49.760 --> 0:01:51.160 so they have a drop dead deadline. 0:01:51.280 --> 0:01:55.040 The twenty twenty three short is not terribly engaging. 0:01:55.160 --> 0:01:58.040 In the first forty three seconds, it feels like a 0:01:58.040 --> 0:02:01.000 product demo, which of course it is. But what can 0:02:01.040 --> 0:02:03.640 we expect from the twenty twenty six version. Will this 0:02:03.760 --> 0:02:06.680 be a moment that has Pixar quaking in their boots. 0:02:06.880 --> 0:02:09.880 Well, you know, if open Ai and their partners in 0:02:09.919 --> 0:02:13.520 this can deliver a feature length hit, the very definition 0:02:13.560 --> 0:02:16.600 of studio could change, you know, especially given their goal, 0:02:17.000 --> 0:02:18.960 which is to make this movie for less than thirty 0:02:19.000 --> 0:02:21.840 million dollars, which is nothing really compared to the hundreds 0:02:21.880 --> 0:02:24.360 of millions it can cost a major studio to make 0:02:24.400 --> 0:02:27.800 a feature length animated film. And then there's the timeline 0:02:27.840 --> 0:02:29.840 that we talked about. You know, open Ai said they 0:02:29.840 --> 0:02:33.360 will wrap in nine months, while Pixar would probably take 0:02:33.639 --> 0:02:35.639 two to four years to make something similar. 0:02:35.880 --> 0:02:38.880 Assuming the film comes out in twenty twenty six and 0:02:38.919 --> 0:02:40.960 it's well received, it can standing. 0:02:40.639 --> 0:02:43.120 Ovation fifteen minutes, and then it. 0:02:43.120 --> 0:02:44.760 Drives real box office. I mean, there are a lot 0:02:44.760 --> 0:02:48.239 of big assumptions. If all of that happens, and it 0:02:48.280 --> 0:02:50.560 happens at a fraction of the cost, at a fraction 0:02:50.600 --> 0:02:52.520 of the time the pix I would do it in 0:02:52.880 --> 0:02:56.120 this would represent kind of a seismic moment for the 0:02:56.120 --> 0:02:56.880 whole of Hollywood. 0:02:57.080 --> 0:03:00.359 It would represent a seismic moment. I think that it's 0:03:00.400 --> 0:03:04.200 something that Hollywood definitely should be worried about. But there 0:03:04.240 --> 0:03:07.400 also are a lot of humans inside this loop. You know, 0:03:07.960 --> 0:03:11.880 OpenAI products like Dolly and Sora will be used for production, 0:03:12.360 --> 0:03:14.520 but humans will be the voice actors, and there are 0:03:14.560 --> 0:03:17.160 some human artists who are even overseeing the project. I 0:03:17.160 --> 0:03:19.720 think most importantly, the screenplay is being written by two 0:03:19.760 --> 0:03:21.800 of the writers from the Paddington franchise. 0:03:22.480 --> 0:03:26.480 How is the industry responding, I'm obviously generative AI was 0:03:26.520 --> 0:03:28.960 one of the huge issues of the writers strike. 0:03:29.240 --> 0:03:30.799 Yeah, and I think there are a lot of people 0:03:30.880 --> 0:03:34.400 outside Hollywood that assume that generative AI is no go 0:03:34.600 --> 0:03:37.960 in the industry, especially because of how tense the strike's got. 0:03:38.560 --> 0:03:41.360 But that's what today's episode is about. I actually got 0:03:41.360 --> 0:03:44.040 to talk to Lila Shapiro, who's a features writer for 0:03:44.080 --> 0:03:47.720 New York Magazine, and her recent piece was titled Hollywood 0:03:47.760 --> 0:03:51.640 already uses generative AI in parenz and is hiding it? 0:03:52.240 --> 0:03:54.440 And in this piece she talks about how generative AI 0:03:54.560 --> 0:03:57.120 is viewed and used in Hollywood. She actually heard from 0:03:57.160 --> 0:04:01.880 studio execs, VFX artists, writers, act directors, and even up 0:04:01.920 --> 0:04:04.800 and coming generative AI studios hoping to work within the 0:04:04.840 --> 0:04:05.640 studio system. 0:04:06.200 --> 0:04:09.440 The victories of the strikes were largely about like what 0:04:09.480 --> 0:04:13.280 people can sort of be forced to do, but people 0:04:13.320 --> 0:04:16.760 are being sort of pressed to do it off the 0:04:16.880 --> 0:04:21.799 record informally by people lower down the production chain because 0:04:21.839 --> 0:04:25.320 everyone is pressed for time, and here are some really 0:04:25.360 --> 0:04:29.560 efficient ways to do things. So it's like there are 0:04:29.600 --> 0:04:33.400 these guidelines, but it's limited to how much it's going 0:04:33.440 --> 0:04:38.239 to really like stave off these greater pressures of needing 0:04:38.279 --> 0:04:43.360 to be more efficient, productive, and deal with tightening budgets 0:04:43.400 --> 0:04:44.560 across the industry. 0:04:45.000 --> 0:04:47.400 I think it's important to remember that AI already plays 0:04:47.400 --> 0:04:50.200 a role in Hollywood, and it has for over a decade. 0:04:50.400 --> 0:04:53.159 Lila told me every studio uses some form of AI. 0:04:53.680 --> 0:04:57.599 The thing being argued over in Hollywood is like generative 0:04:57.680 --> 0:05:00.719 art like purely creative task that can take the artistry 0:05:00.720 --> 0:05:03.239 out of a job or threatened to replace jobs entirely. 0:05:03.800 --> 0:05:05.840 But there are a million uses for AI that are 0:05:05.880 --> 0:05:09.200 just replacing gruntwork. I mean, the writer's strike resulted in 0:05:09.200 --> 0:05:11.719 a contract that says scripts can't be generated by AI, 0:05:12.040 --> 0:05:14.680 but if a writer wants to use AI, they can 0:05:14.839 --> 0:05:15.120 use it. 0:05:15.440 --> 0:05:16.120 And this is how. 0:05:16.040 --> 0:05:18.840 AI is treated in most parts of the industry. Here's 0:05:18.920 --> 0:05:19.520 Lila again. 0:05:19.839 --> 0:05:25.240 Studios aren't advertising what they're doing with it because the 0:05:25.279 --> 0:05:28.839 public response is so negative and there aren't like tons 0:05:28.880 --> 0:05:31.960 of examples that we know about. To give you an example, 0:05:32.160 --> 0:05:36.279 like you've looked at what happened with The Brutalist, where 0:05:36.360 --> 0:05:40.800 the director came under criticism when it developed that they'd 0:05:40.920 --> 0:05:45.760 used generative AI to adjust the audio on some of 0:05:45.800 --> 0:05:47.680 the dialogue, you know, And I think it had to 0:05:47.720 --> 0:05:52.840 do with adjusting the actors like Hungarian accent, and so 0:05:52.880 --> 0:05:56.599 something like that I think is extremely widespread, and it's 0:05:56.680 --> 0:05:59.440 almost it's hard to argue with. You're like, well, there's 0:05:59.480 --> 0:06:03.680 not something sacred about like adjusting audio levels in a 0:06:03.720 --> 0:06:07.320 certain way, and now there's this technology that makes it 0:06:07.440 --> 0:06:10.520 much easier and faster. If you look at like specific 0:06:10.640 --> 0:06:14.159 little windows. You're going to find everyone is doing it, 0:06:14.240 --> 0:06:15.480 and why wouldn't they. 0:06:15.920 --> 0:06:18.839 But here's the thing. The only people motivated to scream 0:06:18.880 --> 0:06:21.520 about it from the rooftops are AI companies like Open 0:06:21.560 --> 0:06:24.400 Ai or other creative studios that are dedicated to the 0:06:24.480 --> 0:06:27.240 use of generative AI, which means that Lila had to 0:06:27.279 --> 0:06:29.960 really dig to understand the true role of AI in 0:06:30.040 --> 0:06:34.120 Hollywood right now, not to mention the industry's collective opinion 0:06:34.160 --> 0:06:34.440 on it. 0:06:34.520 --> 0:06:35.800 So what you're saying is we're going to get an 0:06:35.839 --> 0:06:36.880 expose today. 0:06:37.120 --> 0:06:40.240 Well, I started my conversation with Lila asking about a 0:06:40.279 --> 0:06:43.479 swanky Hollywood party she attended. There were a ton of 0:06:43.520 --> 0:06:45.960 high powered people in attendance, but so many of the 0:06:46.000 --> 0:06:49.360 people Lilah interviewed there refused to be named. And that's 0:06:49.400 --> 0:06:51.880 because the party was a launch party for a new 0:06:52.040 --> 0:06:55.000 generative AI studio. I'll let Lila take it from here. 0:06:55.800 --> 0:07:00.800 So the party was put on by one of these 0:07:00.960 --> 0:07:06.279 new AI studios. This one is called Asteria, and it 0:07:06.360 --> 0:07:09.400 had described itself as sort of the Pixar of AI. 0:07:10.560 --> 0:07:13.960 And part of the reason the studio had some buzz 0:07:14.000 --> 0:07:19.320 around it was that it was run partly by the 0:07:19.400 --> 0:07:23.480 actress and showrunner and writer Natasha Leone who sort of 0:07:23.520 --> 0:07:28.040 this like Indie World Darling. And the other component of 0:07:28.080 --> 0:07:31.520 it is that this studio is sort of positioning themselves 0:07:31.680 --> 0:07:36.800 as the ethical AI studio in that they are training 0:07:36.840 --> 0:07:41.760 their model, they say, only on licensed material. So there's 0:07:41.800 --> 0:07:45.000 this sense that a lot of artists feel, rightly so 0:07:45.080 --> 0:07:49.360 that like their work is being scraped, they're not being compensated, 0:07:49.400 --> 0:07:52.920 and now they're being replaced by these models that can 0:07:53.000 --> 0:07:56.960 do what they did much more efficiently. So Asteria is 0:07:57.000 --> 0:08:00.640 trying to sort of get around that problem by saying 0:08:00.640 --> 0:08:03.760 that they're licensing all their footage and they're building a 0:08:03.800 --> 0:08:06.480 model with a group of engineers who kind of come 0:08:06.520 --> 0:08:09.920 out of the AI scene in the Silicon Valley. This 0:08:10.000 --> 0:08:12.360 party that I went to was the launch party for 0:08:12.560 --> 0:08:16.640 their model, and it's in this big room and there 0:08:16.640 --> 0:08:21.040 are these images that are created by their model projected 0:08:21.080 --> 0:08:26.080 across the walls, trippy things of like a cloud turning 0:08:26.080 --> 0:08:29.120 into a man and the man falling into the clouds 0:08:29.120 --> 0:08:32.440 and turning back into them, or like undulating fields and 0:08:32.520 --> 0:08:33.680 galloping horses. 0:08:33.960 --> 0:08:35.600 Everything is very dolliesque. 0:08:35.840 --> 0:08:38.440 Yes, there was no presentation or anything, so you don't 0:08:38.480 --> 0:08:41.920 really know how they made it or what the process 0:08:42.200 --> 0:08:47.600 was of producing those moving images. It was very like scene. 0:08:48.160 --> 0:08:51.000 They weren't leading with an explanation of sort of what 0:08:51.040 --> 0:08:54.360 they were and what they were about and how they 0:08:54.360 --> 0:08:55.520 were doing these things. 0:08:55.760 --> 0:08:58.440 There was sort of a weird thing that you described 0:08:58.480 --> 0:09:00.240 in your article that there were a lot of people 0:09:00.240 --> 0:09:01.560 who didn't want to give their names. 0:09:01.920 --> 0:09:04.240 What was that about, right? 0:09:05.000 --> 0:09:08.120 I think that people still feel that there is a 0:09:08.240 --> 0:09:14.000 kind of stigma like associated with using generative AI. When 0:09:14.000 --> 0:09:17.800 the article came out, a lot of people were really 0:09:17.840 --> 0:09:20.880 upset with Natasha because they're like, this is a betrayal. 0:09:21.480 --> 0:09:23.800 So people would tell me, like, I think this is 0:09:23.840 --> 0:09:26.479 going to like alienate me from my friends and colleagues 0:09:26.520 --> 0:09:29.720 if they know that I'm here and then I'm interested 0:09:29.800 --> 0:09:32.839 and I'm looking into it. Even though it's becoming increasingly 0:09:32.880 --> 0:09:37.280 widespread and it's happening everywhere, people still don't like it. 0:09:38.040 --> 0:09:41.320 I understand that there's still a stigma around the use 0:09:41.320 --> 0:09:44.760 of AI and supporting companies that use AI, even though 0:09:44.880 --> 0:09:46.440 every company now uses AI. 0:09:46.840 --> 0:09:50.000 Yeah, a lot of what's being done with generative AI 0:09:50.600 --> 0:09:56.320 is replacing what CGI was, and it's not like there's 0:09:56.360 --> 0:10:00.680 something that's sacred about CGI or like rotoscopic where you're 0:10:00.720 --> 0:10:06.960 like very painstakingly cutting out a person's image and moving 0:10:06.960 --> 0:10:12.520 them into another background. That's not like a pleasurable activity. 0:10:12.760 --> 0:10:16.640 It's not a creative activity. It's like a roat task 0:10:16.800 --> 0:10:18.640 that has to be done over and over and over 0:10:18.679 --> 0:10:21.560 and over and over again. If you have a way 0:10:21.600 --> 0:10:25.840 to do that where it happens much faster, saving you 0:10:25.880 --> 0:10:30.239 like days or weeks of time and looks the same essentially, 0:10:30.600 --> 0:10:33.000 then like why wouldn't you want to do that? A 0:10:33.040 --> 0:10:35.560 lot of the people who are most interested in experimenting 0:10:35.600 --> 0:10:38.960 with it are like CGI VFX people because they're like 0:10:39.040 --> 0:10:42.720 tech people who are interested in emerging technologies, and their 0:10:42.800 --> 0:10:46.760 work itself is part was once in an emerging technology, 0:10:46.840 --> 0:10:48.960 so it's not like they're like dead set against it, 0:10:49.040 --> 0:10:52.120 but it does have the effect of no matter what 0:10:52.160 --> 0:10:56.360 way you sort of slice it, visual effects is now 0:10:56.679 --> 0:10:59.679 going to be much faster, which means from like a 0:10:59.720 --> 0:11:02.360 labor perspective, it takes less people to do it, and 0:11:02.360 --> 0:11:05.320 it takes those people a shorter amount of time. And 0:11:05.360 --> 0:11:07.760 that even leaves aside the ethical issue of is this 0:11:07.840 --> 0:11:10.880 based on stolen work? Which hysteria is trying to get 0:11:10.920 --> 0:11:12.920 around with their system. 0:11:13.559 --> 0:11:16.360 So one of the studios you spoke to is Runway. 0:11:16.400 --> 0:11:18.640 Can you tell me a little bit more about them, 0:11:18.679 --> 0:11:21.600 and like what do they do exactly. 0:11:23.080 --> 0:11:29.840 Runway is this company that they've designed a model that 0:11:29.960 --> 0:11:35.640 generates video and they also, you know, have the ambitions 0:11:35.679 --> 0:11:39.080 to become a studio themselves. And they've now produced a 0:11:39.120 --> 0:11:42.560 pilot and they run these film festivals that are AI 0:11:42.640 --> 0:11:47.840 film festivals. So they're really embedded in Hollywood at this point, 0:11:47.880 --> 0:11:52.360 and they've gotten a lot further there than the big 0:11:52.640 --> 0:11:56.840 tech companies have around. When I started reporting, I think 0:11:56.920 --> 0:12:01.360 the picture had been big tech doesn't know how to 0:12:01.440 --> 0:12:05.360 operate in Hollywood, whereas I think Runway came in and 0:12:05.440 --> 0:12:09.840 started having meetings, and at this point they're meeting with 0:12:09.920 --> 0:12:14.760 every studio. They had the first publicly announced partnership with 0:12:14.920 --> 0:12:18.679 the studio, which was Lionsgate. But they have a model 0:12:19.040 --> 0:12:23.960 which can do incredible special effects. When I was talking 0:12:24.040 --> 0:12:28.600 to the Lionsgate executive and other filmmakers who are working 0:12:28.600 --> 0:12:32.640 with Runway, they talked a lot about how, say you 0:12:32.679 --> 0:12:37.240 want to have a shot of ten thousand soldiers in 0:12:37.280 --> 0:12:41.600 the mountains during a snowstorm. To really shoot it, they'd 0:12:41.600 --> 0:12:43.640 have to go to the Himalayas, and it would take 0:12:43.679 --> 0:12:46.640 three days and cost many millions of dollars. And then 0:12:47.000 --> 0:12:51.959 using Runway, they can just create that for ten thousand dollars. 0:12:52.040 --> 0:12:56.360 And it can also do like much more insignificant things 0:12:56.679 --> 0:12:59.360 like say you have a shot, but you want an 0:12:59.400 --> 0:13:04.840 extra foot of image around every part of the scene. 0:13:05.200 --> 0:13:06.880 We can do that. So it can do a lot 0:13:06.920 --> 0:13:12.200 of things with varying degrees of excellence, all of which 0:13:12.240 --> 0:13:17.000 are much faster than alternative ways of doing those things. 0:13:17.600 --> 0:13:20.040 In your piece, there was one paragraph that really struck me. 0:13:20.120 --> 0:13:23.200 It's a paraphrase of something a lions Gate exact told you. 0:13:23.640 --> 0:13:26.160 With a library as large as lions Gates, they could 0:13:26.240 --> 0:13:29.880 use Runway to repackage and resell what the studio already owned. 0:13:30.400 --> 0:13:33.120 You know, I already feel like we're in. 0:13:33.000 --> 0:13:35.760 The We have sequelitis. 0:13:35.160 --> 0:13:39.720 Right now in Hollywood, and I think people who love movies, 0:13:40.600 --> 0:13:44.000 like myself are often lamenting the fact that everything seems 0:13:44.040 --> 0:13:47.520 to be something that already existed. I guess my question 0:13:47.600 --> 0:13:50.960 for you is, do you feel like we are increasingly 0:13:51.080 --> 0:13:56.239 just pushing into repackaged, repurpose content territory that is Hollywood 0:13:56.360 --> 0:14:00.480 or is that lament a little bit pollyannish. 0:14:00.440 --> 0:14:03.880 Like, of course you hear something like that and it's like, 0:14:03.960 --> 0:14:08.520 that's depressing. Now, I'd say the alternate argument to that, 0:14:08.640 --> 0:14:11.040 which I heard a lot from the people at Hysteria, 0:14:11.240 --> 0:14:16.000 certainly Natasha Leone, is that sure, that's one possibility. Another 0:14:16.120 --> 0:14:21.080 possibility is that this technology levels the playing field right, 0:14:21.280 --> 0:14:25.600 fresh voices, new people who would otherwise be unable to 0:14:27.040 --> 0:14:30.040 get a foothold in Hollywood will be in a position 0:14:30.120 --> 0:14:34.920 to make films. And so there are certainly people who 0:14:35.000 --> 0:14:38.840 are arguing that this can lead to a new golden 0:14:38.960 --> 0:14:44.240 era of independent filmmaking. I don't think that's impossible, but 0:14:44.640 --> 0:14:50.200 the worst case scenario doesn't seem that improbable either, Which 0:14:50.240 --> 0:14:53.000 is what which is that like it'll just be all 0:14:53.080 --> 0:14:55.800 slop from here on out. The slop will just get 0:14:55.840 --> 0:14:58.840 more and more and more slop, Like because we're so 0:14:58.960 --> 0:15:03.800 focused in like chure out content and if someone can 0:15:04.120 --> 0:15:07.280 just click a button and star in like seven seasons 0:15:07.440 --> 0:15:10.440 of Game of Throne knockoff, they're going to want to 0:15:10.520 --> 0:15:17.720 do that. What are the incentives for making quality content? 0:15:19.120 --> 0:15:21.040 But at the same time, like, I don't think it's 0:15:21.080 --> 0:15:23.240 an all or nothing. I'm sure it'll just continue to 0:15:23.280 --> 0:15:24.480 be kind of both. 0:15:25.080 --> 0:15:28.640 It's a matter of if people who care about movies 0:15:29.120 --> 0:15:33.200 and the art form of movies will invite these kinds 0:15:33.240 --> 0:15:35.640 of movies into their lives or worlds. 0:15:35.920 --> 0:15:38.560 Yeah. One thing I thought was really interesting in response 0:15:38.600 --> 0:15:40.560 to the piece is at the end of the piece, 0:15:40.800 --> 0:15:45.160 Natasha talks about how her neighbor was David Lynch before 0:15:45.200 --> 0:15:49.080 he died, and she related to me like this conversation 0:15:49.240 --> 0:15:51.920 she had with him about asking him his thoughts, and 0:15:52.600 --> 0:15:55.280 his response was to pick up a pencil and say 0:15:55.440 --> 0:15:58.920 this is AI, basically like, it's a tool. What matters 0:15:59.040 --> 0:16:01.560 is how you use it. And then all these people 0:16:01.640 --> 0:16:05.160 online were like, how dare she drag David Lynch's name 0:16:05.600 --> 0:16:08.160 through the mud like this? He would never have used AI, 0:16:08.560 --> 0:16:11.880 But I thought it was actually very probable that he 0:16:11.880 --> 0:16:14.760 would have said that because he was a constantly evolving 0:16:15.200 --> 0:16:19.600 filmmaker who did use many tools that were at the 0:16:19.640 --> 0:16:23.680 time that he began using them controversial and something that 0:16:23.760 --> 0:16:26.480 others were saying would lead to the death of art 0:16:26.680 --> 0:16:30.520 and the death of filmmaking. So I do think it's 0:16:30.560 --> 0:16:33.800 true that it's just a tool and that like, of course, 0:16:33.840 --> 0:16:36.920 people will make interesting things with it, and there are 0:16:37.080 --> 0:16:40.200 interesting artists right now who are experimenting with it, and 0:16:40.440 --> 0:16:44.040 we haven't seen their work yet, but I think once 0:16:44.160 --> 0:16:46.880 those projects start to come out, there will be a 0:16:46.960 --> 0:16:51.280 shift in attitude. I think the Netflix. Ted Surrando said this, 0:16:51.400 --> 0:16:53.000 like he thinks that it's not just going to make 0:16:53.040 --> 0:16:56.200 movies much cheaper to make, but also ten percent better, 0:16:57.160 --> 0:17:00.480 and that the reason it would make movies ten percent 0:17:00.520 --> 0:17:03.080 better is because like, oh, we'll be more freed up 0:17:03.080 --> 0:17:07.800 to spend more time on the creativity. And to me, 0:17:07.880 --> 0:17:11.080 it's like, well, that's certainly not an inevitable outcome. It 0:17:11.160 --> 0:17:14.000 could easily be just that it's faster and now you're 0:17:14.040 --> 0:17:16.359 expected to just churn out more and more and more 0:17:16.400 --> 0:17:20.639 and more, and you're never giving that time back to 0:17:20.840 --> 0:17:25.120 invest in the creativity. 0:17:30.320 --> 0:17:34.320 After the break, Will AI ever completely replace human filmmakers? 0:17:34.760 --> 0:17:46.880 Stay with us? 0:17:51.240 --> 0:17:54.639 I want to pivot a little bit here to the 0:17:55.040 --> 0:17:58.600 thing that people are focused on most in any conversation 0:17:58.680 --> 0:18:04.160 about AI, which is job. It's no secret that creative 0:18:04.200 --> 0:18:07.480 workers in Hollywood are struggling. Whether it be writers, actors, 0:18:08.560 --> 0:18:10.639 I don't know. It seems like there's a lot on 0:18:10.680 --> 0:18:14.000 the line right now. What were people saying to you 0:18:14.280 --> 0:18:17.080 in your piece, and in your reporting about the future 0:18:17.200 --> 0:18:20.440 of the quote unquote Hollywood workforce. 0:18:21.040 --> 0:18:24.240 Yeah, I mean, I think there's no question that there 0:18:25.160 --> 0:18:29.600 has already been job loss and that there will probably 0:18:29.640 --> 0:18:32.440 be more job loss. It's very hard, though, to really 0:18:32.480 --> 0:18:34.919 pin down the reasons why. When I was talking to 0:18:35.080 --> 0:18:38.360 unions in Hollywood for the piece, no one was sort 0:18:38.359 --> 0:18:41.160 of able or willing to say, like, our members lost 0:18:41.160 --> 0:18:45.680 to this many jobs because of AI. There's also problems 0:18:45.720 --> 0:18:50.399 like the box office is down and budgets are tighter. 0:18:51.040 --> 0:18:53.439 But all of that said, I think there are some 0:18:53.560 --> 0:18:58.400 specific areas where it's pretty clear that like AI can 0:18:58.560 --> 0:19:01.439 do as good job as a person, which is not 0:19:01.560 --> 0:19:05.960 everywhere by any means. But for example, the junior studio executive, 0:19:06.280 --> 0:19:09.800 a big part of that person's job was to write 0:19:09.920 --> 0:19:15.159 script analyzes, and that's something that generative AI can do 0:19:15.200 --> 0:19:17.560 a really good job of and it's not really a 0:19:17.600 --> 0:19:21.480 creative enterprise exactly. When the studio executive talked to me 0:19:21.520 --> 0:19:24.680 about it, that was how she came up writing script 0:19:24.720 --> 0:19:28.920 summaries for her bosses, and now she doesn't need an 0:19:29.000 --> 0:19:31.600 assistant to do that for her. But at the same 0:19:31.640 --> 0:19:33.560 time she was like, but I am worried about what 0:19:33.640 --> 0:19:36.680 is the next generation of studio executives. How are they 0:19:36.720 --> 0:19:40.960 going to learn when this core function of their job 0:19:41.040 --> 0:19:46.600 is no longer necessary? And so who am I training now? 0:19:46.760 --> 0:19:47.120 No one. 0:19:48.200 --> 0:19:51.960 Is it fair to fear that AI will take over 0:19:52.040 --> 0:19:53.320 filmmaking altogether. 0:19:54.000 --> 0:19:57.560 I think that's a reasonable thing to be somewhat afraid of, 0:19:57.800 --> 0:20:02.800 although I do think that it's a pretty complicated question actually, 0:20:02.880 --> 0:20:07.080 because do I think that people will lose the urge 0:20:07.119 --> 0:20:12.159 to create art? No, But it could be that we 0:20:12.280 --> 0:20:16.159 reach a time where there's no economic incentive to do 0:20:16.280 --> 0:20:23.679 it because most people spend their time scrolling TikTok, and 0:20:23.720 --> 0:20:29.040 that there's no more like eyeballs that want to engage 0:20:29.160 --> 0:20:31.600 in serious art. And I mean that's kind of true 0:20:31.640 --> 0:20:37.639 for not just Hollywood, but also publishing, writing, podcasting, everything. 0:20:37.920 --> 0:20:41.840 Is there going to be money to keep making original work? 0:20:42.119 --> 0:20:45.000 But I don't think that we're in a place where 0:20:45.680 --> 0:20:50.359 AI could produce a good movie without human involvement. And 0:20:50.440 --> 0:20:52.080 I don't really know that we will ever be in 0:20:52.119 --> 0:20:54.280 a place where AI could make a good movie without 0:20:54.359 --> 0:20:57.000 human involvement. And core feature of AI is that it 0:20:57.040 --> 0:21:00.199 has no taste. I think sometimes when you have these 0:21:00.240 --> 0:21:03.719 conversations with people who are like thinking about this vision 0:21:04.080 --> 0:21:09.240 where like AI is replacing everything. It's predicated on the 0:21:09.480 --> 0:21:11.680 idea that there's going to continue to be like an 0:21:11.720 --> 0:21:14.840 exponential leap forward and what it can do. And like, 0:21:14.960 --> 0:21:18.840 right now, like AI can't write a good script, for instance, 0:21:19.160 --> 0:21:21.359 and nobody thinks it can. I mean, that's part of 0:21:21.440 --> 0:21:27.920 why the writer's victory during the strikes wasn't that meaningful, 0:21:28.880 --> 0:21:31.920 because right now it really can't create a. 0:21:31.840 --> 0:21:34.200 Good script, but it could in the future. 0:21:34.640 --> 0:21:39.160 Maybe it seems like a real question mark. Meaningful artwork 0:21:39.720 --> 0:21:46.160 has to reflect human choice. So you can imagine someone 0:21:46.480 --> 0:21:50.760 using chat GBT as part of their writing process, but 0:21:50.840 --> 0:21:53.920 it would still have to be led by a person. 0:21:54.440 --> 0:21:58.919 Because I don't think that AI can produce meaningful It 0:21:58.920 --> 0:22:01.840 doesn't have any motivation, It doesn't have any desire to 0:22:01.960 --> 0:22:03.240 express anything. 0:22:03.440 --> 0:22:05.960 Right, And that's why taste seems to be the final 0:22:05.960 --> 0:22:09.359 frontier that would need to be entirely disrupted, which is 0:22:09.880 --> 0:22:11.840 from where I stand, seemingly impossible. 0:22:12.680 --> 0:22:16.400 Yeah, it's just like a different order of business entirely. 0:22:17.040 --> 0:22:18.560 A lot of people that I talk to in AI 0:22:19.200 --> 0:22:23.040 don't think that that is going to happen. They more 0:22:23.160 --> 0:22:26.560 think that it will in the realm of Hollywood, say, 0:22:26.840 --> 0:22:30.520 get better and better and better at doing these specific 0:22:30.640 --> 0:22:35.560 technical functions, which is so different than give me seven 0:22:35.640 --> 0:22:37.960 seasons of whatever TV show. 0:22:38.560 --> 0:22:41.520 Did you get a sense at all in reporting this 0:22:41.680 --> 0:22:44.879 that of like how lay people feel about more and 0:22:44.960 --> 0:22:47.280 more AI content coming out of Hollywood? 0:22:47.400 --> 0:22:49.960 I mean most people, I don't think feel great about it. 0:22:50.440 --> 0:22:53.640 There is, like, I think, a shift since the writer strikes, 0:22:53.840 --> 0:22:58.000 of just more and more acceptance, partly also because chat 0:22:58.119 --> 0:23:02.480 GPT is so widely used now and so more and 0:23:02.520 --> 0:23:05.360 more people are having some form of interaction with it 0:23:05.440 --> 0:23:07.440 and being like this can be helpful. Oh, I could 0:23:07.480 --> 0:23:09.680 do this thing, that would be helpful. So I think 0:23:09.720 --> 0:23:11.920 a big message right now is like, if you want 0:23:11.920 --> 0:23:14.879 to be employed, then you have to use AI. And 0:23:14.920 --> 0:23:18.960 that's true for like every industry right now, not just Hollywood. 0:23:19.359 --> 0:23:23.280 It's like, if you take away the environmental problems, if 0:23:23.320 --> 0:23:27.479 you take away the theft of the world's artwork and 0:23:27.680 --> 0:23:30.280 set those issues aside for a second. 0:23:30.119 --> 0:23:32.960 Those two small factors, those. 0:23:32.800 --> 0:23:37.800 Two small factors, then I do think that it is 0:23:37.840 --> 0:23:41.239 a tool, and there's nothing inherently frightening about it, and 0:23:41.280 --> 0:23:45.000 I think there are certain areas where it's just here 0:23:45.080 --> 0:23:47.840 to stay. So it's not that useful to be like 0:23:47.960 --> 0:23:52.119 that's bad, because it's just the nature of the world 0:23:52.280 --> 0:23:54.919 right now. So I think some of the reactions to 0:23:55.000 --> 0:23:57.320 it don't make a ton of sense. Like I don't 0:23:57.320 --> 0:24:02.679 think that using AI currently curdles the work, because I 0:24:02.720 --> 0:24:05.080 think there's tons of ways to use it that the 0:24:05.200 --> 0:24:08.160