Ben Affleck & Matt Damon on The Limits of AI in Movie Making (YouTube Video Transcript)

(00:00:00) Your YouTube transcript will appear here (00:00:01) the Joe Rogan experience. (00:00:03) >> A lot of the stuff that was going on (00:00:04) with the strikes was centered around AI (00:00:07) and what AI is going to do to the (00:00:09) business. Like what where do you feel is (00:00:12) going to be like the biggest problem (00:00:14) with AI? Is it going to be with people's (00:00:16) likenesses? Because there's a lot of (00:00:18) that where they want they want to use (00:00:19) extras and own their digital rights (00:00:21) forever essentially be able to recreate (00:00:23) them in any kind of film. But then (00:00:25) there's also you're going to have films (00:00:27) that are written by artificial (00:00:29) intelligence. You're going to have (00:00:31) scenes that don't involve people. And it (00:00:35) gets weird, right? (00:00:36) >> It gets really weird, but there's (00:00:38) actually an area for him. (00:00:39) >> Yeah. We've been spending time looking (00:00:41) at this. Like my belief is it's sort of (00:00:42) like what's going to happen with (00:00:43) electricity? (00:00:44) >> Well, a lot of shit's going to happen (00:00:45) with electricity. Some of it's going to (00:00:47) be good. Some of it's going to change (00:00:48) stuff. Some of it's going to be like, (00:00:50) you know, this is going to be, you know, (00:00:52) [ __ ] that kills a bunch of people. like (00:00:53) it's it's it's opening a door that you (00:00:55) can't um you know say well talk about in (00:00:58) a kind of a blanket way but I think with (00:01:00) what I see is like for example if you (00:01:02) try to get chat GBT or Claude or Gemini (00:01:06) to write you something it's really (00:01:08) shitty and it's shitty because by its (00:01:10) nature it goes to the mean to the (00:01:12) average and it's and it's not reliable (00:01:16) and it's I mean I just can't even stand (00:01:17) to see what writes now it's a useful (00:01:19) tool if you're a writer and you're going (00:01:21) h what's the I'm trying to set something (00:01:23) up where somebody sends someone a letter (00:01:25) but it's delayed two days and gets and (00:01:27) it can give you some examples of that. I (00:01:29) actually don't think it's very likely (00:01:31) that it can it's going to be able to (00:01:33) write anything meaningful or and in (00:01:35) particular that it's going to be making (00:01:38) movies like from whole cloth like Tilly (00:01:39) Nor like that's [ __ ] I don't think (00:01:41) that's going to happen. I think it's not (00:01:43) I think it actually it turns out the (00:01:44) technology is not progressing in exactly (00:01:46) the same way they sort of presented it. (00:01:48) Um, and really what it is is going to be (00:01:50) a tool just like sort of visual effects (00:01:53) and yeah, it needs to have language (00:01:54) around it. You need to protect your name (00:01:55) and likeness. You can do that. You can (00:01:57) watermark it. Your those laws already (00:01:59) exist. You can't I can't sell your (00:02:01) [ __ ] picture for money. I can't. You (00:02:02) can sue me. Period. I might have the (00:02:04) ability to draw you to make you in a (00:02:07) very realistic way, but that's already (00:02:09) against the law. And the unions are (00:02:12) going to I think the guilds are going to (00:02:13) manage this where it's like okay look if (00:02:16) this is a tool that actually helps us (00:02:17) for example we don't have to go to the (00:02:18) North Pole right we can shoot the scene (00:02:20) here in our parkas and you know whatever (00:02:23) it is and but then make it appear very (00:02:26) realistically as if we're in the North (00:02:27) Pole save us a lot of money a lot of (00:02:29) time we're going to focus on the (00:02:31) performances and not be freezing our ass (00:02:32) up out there and running back inside (00:02:35) that's useful just like Spencer Tracy (00:02:37) and Katherine Heppern used to be like (00:02:39) driving their car and there's a wind (00:02:40) blowing a painting behind them and look (00:02:42) goofy and you know now you know in (00:02:44) computer people use a lot of (00:02:45) computerenerated stuff and some of it is (00:02:48) going to replace just that like instead (00:02:49) of uh 500 guys in Singapore you know (00:02:51) making $2 an hour to to render all the (00:02:54) graphics for a superhero movie there's (00:02:57) going to be able to do that a lot easier (00:02:59) there's already laws around and guild (00:03:01) guidelines around like how many union (00:03:02) extras you have to use but also we've (00:03:05) been tiling extras like there weren't a (00:03:07) million orcs in Middle Earth you know (00:03:09) what I mean there aren't Invictus, there (00:03:10) weren't all those people in the stadium. (00:03:12) Like that's something we've been doing. (00:03:14) It kind of feels to me like the thing we (00:03:16) were talking about earlier where there's (00:03:17) a lot more fear because we have the (00:03:19) sense this existential dread. It's going (00:03:21) to wipe everything out. (00:03:22) >> But that actually runs counter in my (00:03:24) view to what history seems to show which (00:03:27) is a adoption is slow. It's incremental. (00:03:31) Um, I think a lot of that rhetoric comes (00:03:33) from people who are trying to justify (00:03:35) valuations around companies where they (00:03:38) go, "We're going to change everything in (00:03:39) two years. There's going to be no more (00:03:40) work." (00:03:41) >> Well, the reason they're saying that is (00:03:42) because they need to ascribe a valuation (00:03:45) for investment that can warrant the (00:03:47) capex spend they're going to make on (00:03:48) these data centers with the argument (00:03:49) that like, oh, you know, as soon as we (00:03:52) do the next model, it's going to scale (00:03:53) up. It's going to be three times as (00:03:54) good. Except that actually chat GP5 (00:03:57) about 25 time percent better than chat (00:04:00) GP4 and costs about four times as much (00:04:03) in the way of electricity and data. So (00:04:05) those say that's like plateauing. The (00:04:08) early AI the line went up very steeply (00:04:11) and it's now sort of leveling off. I (00:04:13) think it's because and yes it'll get (00:04:15) better but it's going to be really (00:04:17) expensive to get better. And a lot of (00:04:19) people were like, "Fuck this. We want (00:04:20) chat GB4." Because it turned out like (00:04:22) the vast majority of people who use AI (00:04:25) are using it to like as like companion (00:04:28) bots to chat with at night and stuff. (00:04:30) There's no work, there's no (00:04:31) productivity, there's no value to it. I (00:04:33) would argue there's also not a lot of (00:04:35) social value to getting people to like (00:04:38) focus on an AI friend who's, you know, (00:04:40) telling you that you're great and (00:04:42) listening to everything you say and (00:04:43) being sick of fantic. But that's sort of (00:04:46) a side issue. think for this particular (00:04:48) purpose like the way I see the (00:04:50) technology and what it's good at and (00:04:51) what it's not it's going to be good at (00:04:53) filling in all the places that are (00:04:54) expensive and burdensome and they make (00:04:56) it harder to do it and it's always going (00:04:58) to rely fundamentally on the human (00:05:00) artistic aspects of it well I think the (00:05:03) more it becomes ubiquitous the more (00:05:05) people are going to appreciate real (00:05:07) things that are made by real people you (00:05:09) know like you're you still appreciate a (00:05:11) handmade table you know you're you're (00:05:13) gonna appreciate like did you see Um, (00:05:16) uh, The Beast in Me, Claire Danes. (00:05:18) >> Yeah. (00:05:18) >> No, I didn't. [ __ ] great. Yeah, I (00:05:20) heard it was great. (00:05:20) >> That lady. Woo. (00:05:22) >> It's terrific. (00:05:23) >> Woo. When she's in a scene, you're just (00:05:25) like, Jesus Christ. Like great. (00:05:27) >> Like you like her [ __ ] lips are (00:05:28) quivering. Like you believe everything (00:05:30) that she's saying. (00:05:31) >> But you're right. People want that. (00:05:35) >> I say like I I did this interview with (00:05:38) uh with Dwayne Johnson because they, you (00:05:39) know, they when people are in these (00:05:41) awards things, they sometimes have other (00:05:43) actors interview them, you know. And I (00:05:45) did this interview with Dwayne and and (00:05:46) and I asked him there's this scene in (00:05:48) the smashing machine where where he's (00:05:51) overdosed on drugs and his buddy comes (00:05:52) to see him in the hospital. (00:05:54) >> Yeah. (00:05:54) >> And and it really walloped me this (00:05:57) scene. I thought it was so great. And (00:05:59) and I asked him and I was just like, (00:06:00) "Can you just tell me about this scene? (00:06:02) Like did Benny Benny Safy directed it? (00:06:04) Did Benny write this write that? Did you (00:06:06) work on that scene with them? Did you?" (00:06:07) He goes, "No, we we actually worked on (00:06:09) it together." And I go, "But how did (00:06:10) that scene come to be?" And Dwayne goes, (00:06:12) "Well, my father was an alcoholic, and I (00:06:15) don't remember if he said substance (00:06:16) abuse or alcoholic, but I didn't know (00:06:18) the man. I don't want to impute him, but (00:06:19) but he had he had a substance issue, (00:06:21) whatever it was." He goes, "And and when (00:06:23) he would talk to me, uh, you know, (00:06:26) that's how he (00:06:28) would defend himself. It was almost a (00:06:29) bargaining thing cuz there's this thing (00:06:31) when this guy comes to him, he's (00:06:32) overdosed, and Dwayne's amazing in this (00:06:35) scene. He's he's going like he's going (00:06:37) like, "Yeah, isn't it crazy?" And then I (00:06:38) woke up and thought, I mean, I could (00:06:39) hear him, but I couldn't really hear (00:06:40) him. and you see him and he's kind of (00:06:41) tap dancing and his friend finally kind (00:06:43) of holds his feet to the fire (00:06:45) >> and at that moment Dwayne (00:06:48) >> literally (00:06:49) starts to burst into tears and just (00:06:51) pulls the hospital sheet up over his (00:06:54) head and it's like and it's and it's I (00:06:57) mean it's just it was I'm I'm not doing (00:07:00) it justice if you haven't I mean I know (00:07:01) you see I know you've seen it (00:07:03) >> but um he said yeah so he explains that (00:07:07) about his father and then he goes (00:07:09) >> and and Uh when my mom was diagnosed (00:07:11) with stage three lung cancer, I was with (00:07:13) her when the oncologist came in and she (00:07:14) was lying in the hospital bed. And when (00:07:17) he gave her the news, she pulled the (00:07:18) sheet up over her head and I looked at (00:07:22) her and she just looked like a little (00:07:23) like a little kid, you know? And I was (00:07:26) like, "All right." Like, so that right (00:07:30) is two traumatic events from this guy's (00:07:33) life, right? From his life experience. (00:07:36) And the actor in him, right, sees this (00:07:39) scene, (00:07:41) goes into his memory, pulls these two (00:07:44) things out, understands that they're (00:07:46) appropriate for this scene, and he can (00:07:47) marry them together in the scene, and (00:07:50) then he goes and performs it that way. (00:07:52) And a dude walking in off the road, goes (00:07:55) to the movies, sees this, understands (00:08:00) somehow that it's [ __ ] real. I I (00:08:03) didn't know why. I that's why I wanted (00:08:04) to ask him how did that scene come to (00:08:06) be. I genuinely didn't know (00:08:08) >> and made me tear up and you know like (00:08:14) that is (00:08:15) >> there's no [ __ ] AI that can do that. (00:08:18) >> It's the whole lot more than (00:08:19) photorealistic images. (00:08:21) >> Yeah. You you could you could you could (00:08:23) have an AI understand Dwayne's face and (00:08:25) move his face into different No [ __ ] (00:08:27) thing could ever do that. the (00:08:29) complications of real life experiences (00:08:31) relayed. (00:08:32) >> That is a completely human. That is an (00:08:34) that is an artist. That's a piece of (00:08:35) art, right? That comes out of a lived (00:08:37) human experience. (00:08:38) >> NFL playoffs, let's go. DraftKings (00:08:40) Sportsbook, an official sports betting (00:08:42) partner of the NFL makes every moment (00:08:44) feel bigger. Postseason games shift (00:08:47) fast. And with DraftKings live betting (00:08:49) options, you can stay right in the (00:08:51) moment. Plus, DraftKings has your back (00:08:53) with early exit protection. If the (00:08:55) player in your eligible NFL prop bet (00:08:57) goes down at any point in the first (00:09:00) half, you still get paid in cash. New (00:09:03) customers bet just $5. And if the bet (00:09:05) wins, you get 300 in bonus bets (00:09:09) instantly. Download the DraftKings (00:09:11) Sportsbook app now and use the code (00:09:13) Rogan. That's code Rogan to turn five (00:09:15) bucks into 300 in bonus bets if your bet (00:09:19) wins. In partnership with DraftKings, (00:09:21) the crown is yours. (00:09:22) >> Gambling problem? Call 1800 gambler. In (00:09:25) New York, call 8778 wire or text (00:09:27) open467-369. (00:09:29) In Connecticut, call 888-789-7777 (00:09:32) or visit ccpg.org. On behalf of Bootill (00:09:35) Casino and Resort in Kansas, passrough (00:09:37) of per wager tax may apply in Illinois. (00:09:38) 21 and over. Age and eligibility varies (00:09:40) by jurisdiction. Void in Ontario. (00:09:42) Restrictions apply. Bet must win to (00:09:43) receive bonus bets which expire in 7 (00:09:45) days. Minimum odds required. For (00:09:47) additional terms and responsible gaming (00:09:48) resources, see dkg.co/audio. (00:09:51) Limited time offer.