1 00:00:02,520 --> 00:00:07,880 Speaker 1: All Zone Media. Hello and welcome to Better Offline as usual. 2 00:00:07,960 --> 00:00:21,600 Speaker 1: I'm your host ed zitron. A few months ago, open 3 00:00:21,640 --> 00:00:24,800 Speaker 1: Ai showed off Sora, a product that can generate videos 4 00:00:24,800 --> 00:00:27,120 Speaker 1: based on a short text prompt, kind of like chat 5 00:00:27,240 --> 00:00:31,800 Speaker 1: ebt does for text or Daali does for images. These videos, 6 00:00:31,840 --> 00:00:34,680 Speaker 1: which are usually no more than sixty seconds long, can 7 00:00:34,760 --> 00:00:38,040 Speaker 1: at times seem impressive until you notice a little detail 8 00:00:38,040 --> 00:00:40,640 Speaker 1: that breaks the entire facade, like in a video where 9 00:00:40,640 --> 00:00:42,559 Speaker 1: a cat wakes up its owner, but the owner's arm 10 00:00:42,600 --> 00:00:44,760 Speaker 1: appears to be part the cushion and the cat's poor 11 00:00:44,880 --> 00:00:48,360 Speaker 1: explodes out of its arm like an ameba. Reactions to 12 00:00:48,479 --> 00:00:51,440 Speaker 1: Sora's Ai generated videos, and indeed the existence of the 13 00:00:51,440 --> 00:00:54,840 Speaker 1: model itself, have ranged from kind of a breathless hype 14 00:00:54,920 --> 00:00:57,840 Speaker 1: to genuine fear that this will be used to replace 15 00:00:57,920 --> 00:01:01,640 Speaker 1: video producers, in that it can create reality adjacent videos 16 00:01:01,680 --> 00:01:04,880 Speaker 1: that for a few seconds kind of seem real, especially 17 00:01:04,959 --> 00:01:07,280 Speaker 1: in the case in some of open Aye's hand pick 18 00:01:07,360 --> 00:01:12,200 Speaker 1: demo videos. Yet even in these handpicked Sora outputs, you'll 19 00:01:12,240 --> 00:01:15,800 Speaker 1: find these weird little things that immediately shatter the illusion, 20 00:01:16,160 --> 00:01:19,119 Speaker 1: like one where a woman's legs awkwardly shuffle, then somehow 21 00:01:19,160 --> 00:01:21,959 Speaker 1: switch sides as she walks around, or blobs of people 22 00:01:22,000 --> 00:01:25,240 Speaker 1: merging in the background of images. These are, on some 23 00:01:25,520 --> 00:01:31,280 Speaker 1: level genuinely remarkable technological achievements, until you consider that what 24 00:01:31,440 --> 00:01:35,560 Speaker 1: they are and what they might do, and that there 25 00:01:35,600 --> 00:01:39,120 Speaker 1: are problems in them that run through the entire fabric 26 00:01:39,200 --> 00:01:43,319 Speaker 1: of artificial intelligence. A little over a month after SAW 27 00:01:43,520 --> 00:01:46,440 Speaker 1: was announced, open AI would debut a series of short films, 28 00:01:46,480 --> 00:01:49,760 Speaker 1: including one called Airhead, where filmmakers Shy Kids told the 29 00:01:49,760 --> 00:01:51,360 Speaker 1: story of a man with a balloon for a head, 30 00:01:51,800 --> 00:01:55,120 Speaker 1: and because this is AI said, balloon changes sizes twenty three, 31 00:01:55,280 --> 00:01:58,240 Speaker 1: twenty four, twenty six, twenty seven, twenty nine, thirty two, 32 00:01:58,280 --> 00:02:01,240 Speaker 1: thirty four, thirty nine, forty one, forty two, forty three, 33 00:02:01,240 --> 00:02:03,960 Speaker 1: and forty five seconds into the piece, at which point 34 00:02:04,000 --> 00:02:06,120 Speaker 1: I stopped counting because it got boring and I really 35 00:02:06,120 --> 00:02:08,280 Speaker 1: don't want to be mean to shy kids, as this 36 00:02:08,600 --> 00:02:12,840 Speaker 1: really isn't their fault. The very nature of filmmaking is 37 00:02:12,840 --> 00:02:15,560 Speaker 1: that you take different shots of the same thing. Something 38 00:02:15,639 --> 00:02:19,200 Speaker 1: that I anticipated SAA was incapable of doing. Is each 39 00:02:19,280 --> 00:02:22,480 Speaker 1: shot is generated fresh a saura itself. Much like all 40 00:02:22,600 --> 00:02:27,360 Speaker 1: generative AI does not actually know anything when one asks 41 00:02:27,560 --> 00:02:29,880 Speaker 1: for a man with a yellow balloon as his head. 42 00:02:30,160 --> 00:02:32,960 Speaker 1: SAURA must then look at the parameters spawn during its 43 00:02:33,000 --> 00:02:36,040 Speaker 1: training process and create an output guessing what a man 44 00:02:36,080 --> 00:02:38,360 Speaker 1: looks like, what a balloon looks like, what a man's 45 00:02:38,480 --> 00:02:41,760 Speaker 1: features are on his body, what color yellow is, what 46 00:02:41,760 --> 00:02:45,680 Speaker 1: the man's doing, and so on and so forth. This 47 00:02:45,800 --> 00:02:49,680 Speaker 1: becomes extremely problematic when you're working in film or television, 48 00:02:49,720 --> 00:02:52,120 Speaker 1: where viewers are far more likely to see when something 49 00:02:52,240 --> 00:02:55,680 Speaker 1: just doesn't look right, a problem exacerbated by moving images, 50 00:02:55,840 --> 00:02:59,800 Speaker 1: high resolution footage, and big television screens which are now ubiquitous. 51 00:03:00,680 --> 00:03:05,400 Speaker 1: Yet the press, as usual, credulously accepted Saura's quote stunning 52 00:03:05,520 --> 00:03:09,360 Speaker 1: videos that were amazing and scary, suggesting to the public 53 00:03:09,400 --> 00:03:11,280 Speaker 1: that we were on the verge of some sort of 54 00:03:11,440 --> 00:03:16,760 Speaker 1: artificial intelligence takeover of the film industry, helping boy Sam Altman, 55 00:03:16,880 --> 00:03:20,440 Speaker 1: their CEO, and his dumbast attempts to convince Hollywood that 56 00:03:20,600 --> 00:03:25,440 Speaker 1: SURRA won't destroy the movie business. These stories only serve 57 00:03:25,520 --> 00:03:28,680 Speaker 1: to help Sam Orman, who desperately needs you to believe 58 00:03:28,680 --> 00:03:31,480 Speaker 1: that Hollywood is scared of Surer and even more scared 59 00:03:31,480 --> 00:03:34,120 Speaker 1: of Generative AI, because the more you talk about fear 60 00:03:34,160 --> 00:03:36,680 Speaker 1: and lost jobs and the machines taking over the less. 61 00:03:36,720 --> 00:03:40,360 Speaker 1: You ask a very very simple question, does any of 62 00:03:40,360 --> 00:03:45,160 Speaker 1: this shit actually work? The answer, it turns out, is 63 00:03:45,200 --> 00:03:48,560 Speaker 1: not very well. In a piece for FX Guide, Mike 64 00:03:48,600 --> 00:03:51,160 Speaker 1: Seymour sat down with Shy Kids, the people behind Airhead, 65 00:03:51,320 --> 00:03:54,720 Speaker 1: and revealed how SORAW is in many ways a little 66 00:03:54,720 --> 00:03:58,680 Speaker 1: bit useless for making films. SAURA takes ten to twenty 67 00:03:58,720 --> 00:04:01,560 Speaker 1: minutes to generate a single three to twenty second shot, 68 00:04:02,000 --> 00:04:04,400 Speaker 1: something that isn't really a problem until you realize that 69 00:04:04,520 --> 00:04:07,520 Speaker 1: until the shot is rendered, you really have absolutely no 70 00:04:07,600 --> 00:04:10,600 Speaker 1: idea what the hell it's going to spit out. Soa 71 00:04:10,880 --> 00:04:13,480 Speaker 1: has no mechanism to connect one shot to another. Even 72 00:04:13,480 --> 00:04:17,640 Speaker 1: with hyperdescriptive prompts. It hallucinates extra features when you haven't 73 00:04:17,680 --> 00:04:20,360 Speaker 1: asked for them. And Shy Kids were shocked by how 74 00:04:20,400 --> 00:04:23,680 Speaker 1: surprised open Ay's researchers were when they requested the ability 75 00:04:23,680 --> 00:04:27,080 Speaker 1: to use a prompt to request a particular angle in 76 00:04:27,120 --> 00:04:31,520 Speaker 1: a shot, a feature that was initially unavailable. It took 77 00:04:32,200 --> 00:04:35,200 Speaker 1: this is what kind of drives me crazy here and 78 00:04:35,240 --> 00:04:38,279 Speaker 1: you'll hear this in the interview with him later. These 79 00:04:38,320 --> 00:04:40,960 Speaker 1: people that are open AI people, and they were making 80 00:04:40,960 --> 00:04:44,560 Speaker 1: this tool for making visual images for making moving images. 81 00:04:44,600 --> 00:04:47,320 Speaker 1: They didn't think that people might want different shots. I'm 82 00:04:47,320 --> 00:04:49,480 Speaker 1: so glad these are the people who were in control 83 00:04:49,520 --> 00:04:53,159 Speaker 1: of the future. Anyway, to quote the piece, it took 84 00:04:53,320 --> 00:04:56,400 Speaker 1: hundreds of generations at ten to twenty seconds a piece 85 00:04:56,440 --> 00:05:00,720 Speaker 1: to make a minute and nineteen second long film. And 86 00:05:00,760 --> 00:05:05,679 Speaker 1: what's really fun about this is that the movie's fine. 87 00:05:05,800 --> 00:05:09,200 Speaker 1: I it was kind of fine. I just I have 88 00:05:09,320 --> 00:05:11,280 Speaker 1: nothing really to say about it. It's a minute and 89 00:05:11,360 --> 00:05:15,720 Speaker 1: twenty seconds long, but it's it kind of works. But also, 90 00:05:15,960 --> 00:05:18,799 Speaker 1: the balloon looks different in every other shot. This isn't 91 00:05:18,880 --> 00:05:23,279 Speaker 1: shy Kids's fault. But also this isn't gonna get better. 92 00:05:23,480 --> 00:05:26,520 Speaker 1: And I will get into why as we go along. 93 00:05:28,080 --> 00:05:31,479 Speaker 1: These tiny little problems I've mentioned, though, they all lead 94 00:05:31,520 --> 00:05:35,359 Speaker 1: to one overwhelming issue that Sora isn't so much a 95 00:05:35,440 --> 00:05:37,800 Speaker 1: tool to make movies as it is a big, fat 96 00:05:37,839 --> 00:05:40,360 Speaker 1: slot machine that spits out footage that may or may 97 00:05:40,400 --> 00:05:43,440 Speaker 1: not be of any use at all. Almost all of 98 00:05:43,440 --> 00:05:47,360 Speaker 1: the footage in Airhead was graded, treated, stabilized, the nutscaled, 99 00:05:48,000 --> 00:05:50,800 Speaker 1: and that ten to twenty second lead time on generations 100 00:05:50,920 --> 00:05:54,520 Speaker 1: was for four hundred and eightp resolution footage, meaning that 101 00:05:54,600 --> 00:05:58,200 Speaker 1: even useful footage needed significant post production work to look 102 00:05:58,200 --> 00:06:00,680 Speaker 1: good enough, and just to give you an idea for 103 00:06:00,760 --> 00:06:02,840 Speaker 1: the non technical members of the audience, and this is fair. 104 00:06:03,839 --> 00:06:06,599 Speaker 1: The video you see on YouTube is usually somewhere between 105 00:06:06,600 --> 00:06:09,920 Speaker 1: seven TWENTYP, ten ADP or four K. The TV shows 106 00:06:09,960 --> 00:06:13,880 Speaker 1: you watch usually ten AP four K or upscale ten ADP. 107 00:06:14,120 --> 00:06:16,359 Speaker 1: These are all lots of numbers. What I'm saying is 108 00:06:16,839 --> 00:06:20,440 Speaker 1: the stuff that SAA spits out, that takes burning a 109 00:06:20,440 --> 00:06:24,680 Speaker 1: small zoo to spit out, is incredibly low resolution. On 110 00:06:24,760 --> 00:06:29,599 Speaker 1: top of not being specific, look to put it as 111 00:06:29,640 --> 00:06:34,119 Speaker 1: plainly as possible, every single time that shy kids wanted 112 00:06:34,120 --> 00:06:37,400 Speaker 1: to generate a shot, even a three second long shot, 113 00:06:37,600 --> 00:06:40,440 Speaker 1: they would give SA a text prompt and then they 114 00:06:40,440 --> 00:06:44,040 Speaker 1: would wait at least ten minutes to find out if 115 00:06:44,080 --> 00:06:47,640 Speaker 1: it was right, and they'd have to accept footage that 116 00:06:47,800 --> 00:06:52,000 Speaker 1: was subprime or inaccurate. And there's a really good example 117 00:06:52,040 --> 00:06:54,479 Speaker 1: of this. If you watch Airhead, a lot of the 118 00:06:54,520 --> 00:06:57,240 Speaker 1: shots are in slow motion, and you may think, no, 119 00:06:57,400 --> 00:07:00,040 Speaker 1: this is a cinematic choice, right, because you kind of 120 00:07:00,160 --> 00:07:02,200 Speaker 1: just admiring this man with a balloon for a head 121 00:07:02,240 --> 00:07:05,880 Speaker 1: going about his business. No, no, no, no no. They 122 00:07:06,000 --> 00:07:08,440 Speaker 1: found that this was just what Sora wanted to give 123 00:07:08,480 --> 00:07:10,880 Speaker 1: them when they asked for it. This was, in and 124 00:07:10,920 --> 00:07:14,520 Speaker 1: of itself a hallucination, in the same way that chat 125 00:07:14,600 --> 00:07:18,560 Speaker 1: GBT will authoritatively tell you that something is true that 126 00:07:18,720 --> 00:07:22,040 Speaker 1: is not sorrow will spit out a man running in 127 00:07:22,080 --> 00:07:27,960 Speaker 1: slow motion despite you not asking for that, And it's 128 00:07:27,960 --> 00:07:31,040 Speaker 1: so weird. They had to quote them do quite a 129 00:07:31,080 --> 00:07:33,880 Speaker 1: bit of adjusting to keep the whole thing from feeling 130 00:07:34,520 --> 00:07:37,920 Speaker 1: like a big slow mode project, and it still kind 131 00:07:37,920 --> 00:07:43,680 Speaker 1: of does. And that's rough. That's really rough. But you know, 132 00:07:43,800 --> 00:07:46,920 Speaker 1: I'm a curious little critter, So I decided to sit 133 00:07:47,000 --> 00:07:49,640 Speaker 1: down with Shy Kids's Walter Woodman to talk about his 134 00:07:49,680 --> 00:07:52,120 Speaker 1: experience with Sora and have him delve a little daper 135 00:07:52,120 --> 00:07:55,040 Speaker 1: into his experience with the product. And I'd say he 136 00:07:55,120 --> 00:07:59,000 Speaker 1: had a far more utopian experience and perspective on the 137 00:07:59,040 --> 00:08:03,560 Speaker 1: whole thing than I excted. Now, some of you might 138 00:08:04,320 --> 00:08:07,320 Speaker 1: critique Walter for being so positive about it, but I 139 00:08:07,320 --> 00:08:09,520 Speaker 1: actually caution you to just listen to what he's saying, 140 00:08:10,040 --> 00:08:13,400 Speaker 1: because Walter's perspective is interesting. He sees this as a tool, 141 00:08:13,440 --> 00:08:15,680 Speaker 1: he doesn't see it as a replacement, and I think 142 00:08:15,680 --> 00:08:18,320 Speaker 1: it's a valid perspective to come at SAA with. I 143 00:08:18,360 --> 00:08:21,560 Speaker 1: also think it's a perspective that kind of accepts a 144 00:08:21,640 --> 00:08:25,440 Speaker 1: conceit of open AI's marketing strategy, that these things will 145 00:08:25,480 --> 00:08:30,520 Speaker 1: get better if they do. Perhaps Walter is right, perhaps 146 00:08:30,560 --> 00:08:33,600 Speaker 1: this will be an essential tool in filmmaking, even though 147 00:08:33,600 --> 00:08:35,440 Speaker 1: he didn't say essential. Don't want to put words in 148 00:08:35,440 --> 00:08:39,240 Speaker 1: the man's mouth, but I don't think that's the case. 149 00:08:40,320 --> 00:08:54,319 Speaker 1: Let me talk to him. You decide for yourself, all right. 150 00:08:54,440 --> 00:08:57,960 Speaker 1: So how did the relationship between Shy Kids and open 151 00:08:58,000 --> 00:08:58,920 Speaker 1: AYE actually begin. 152 00:09:00,160 --> 00:09:03,840 Speaker 2: The relationship between Shy Kids and Open AI began when 153 00:09:03,880 --> 00:09:08,079 Speaker 2: we made an installation for a film called dolly Land, 154 00:09:08,240 --> 00:09:12,560 Speaker 2: which was premiering at Toronto International Film Festival, and we 155 00:09:12,559 --> 00:09:15,480 Speaker 2: were the only people that our friends at Pressman Film 156 00:09:15,600 --> 00:09:19,720 Speaker 2: knew in Toronto, and so we made an installation that 157 00:09:19,840 --> 00:09:26,040 Speaker 2: looked like Salvador Dali's like studio inside of the basement 158 00:09:26,240 --> 00:09:29,679 Speaker 2: of the Saint Regis, which is where he lived and 159 00:09:30,240 --> 00:09:33,880 Speaker 2: made work out of, And inside of that installation we 160 00:09:34,600 --> 00:09:38,360 Speaker 2: made a like you could make your own surrealist painting, 161 00:09:39,520 --> 00:09:41,800 Speaker 2: and the way that you could make that was using 162 00:09:41,880 --> 00:09:48,160 Speaker 2: DOLLI the Open AI program, and so the open AI 163 00:09:48,320 --> 00:09:53,840 Speaker 2: people came to visit and check out the like what 164 00:09:53,920 --> 00:09:56,080 Speaker 2: we were working on, and making sure that it was 165 00:09:56,160 --> 00:09:58,080 Speaker 2: like something that they wanted to be a part of. 166 00:09:58,840 --> 00:09:59,840 Speaker 3: And so. 167 00:10:01,240 --> 00:10:05,440 Speaker 2: They met our producer Sydney, who they loved. She's easy 168 00:10:05,480 --> 00:10:06,120 Speaker 2: to love. 169 00:10:06,360 --> 00:10:07,840 Speaker 3: And they. 170 00:10:09,120 --> 00:10:11,840 Speaker 2: We sent them our previous work and so from there 171 00:10:12,120 --> 00:10:16,520 Speaker 2: they asked us to join this artist group. And then 172 00:10:16,720 --> 00:10:18,800 Speaker 2: when Sora came out, we saw it at the same 173 00:10:18,840 --> 00:10:24,720 Speaker 2: time as everyone else and we yeah, we got tapped 174 00:10:24,720 --> 00:10:27,720 Speaker 2: on the shoulder and said, hey, would you like to 175 00:10:27,800 --> 00:10:29,400 Speaker 2: check this out and try this out? And we said, 176 00:10:29,440 --> 00:10:32,360 Speaker 2: of course, that's how it came to be. 177 00:10:33,280 --> 00:10:37,119 Speaker 1: So how did you on board? Were you just given access? 178 00:10:37,280 --> 00:10:39,959 Speaker 1: Did they give you instructions? Did they physically come to you? 179 00:10:40,480 --> 00:10:44,199 Speaker 2: What was that like it was a top secret. They 180 00:10:44,240 --> 00:10:48,720 Speaker 2: gave us a briefcase and in a cloudy room. 181 00:10:48,960 --> 00:10:49,720 Speaker 3: No, it was. 182 00:10:50,840 --> 00:10:54,000 Speaker 2: Yeah, there was a very simple onboarding process where they 183 00:10:54,080 --> 00:10:58,080 Speaker 2: walked us through the technology as well as some of 184 00:10:58,120 --> 00:11:05,160 Speaker 2: its features, and yeah, it was pretty. It was pretty. 185 00:11:05,400 --> 00:11:07,640 Speaker 2: And then from there they gave us access to begin 186 00:11:08,280 --> 00:11:09,959 Speaker 2: using it and making. 187 00:11:09,600 --> 00:11:13,160 Speaker 1: Things and you were allowed to use it without their presence. 188 00:11:13,200 --> 00:11:14,319 Speaker 1: You had direct access. 189 00:11:14,360 --> 00:11:15,400 Speaker 3: Yep, yep. 190 00:11:16,320 --> 00:11:20,240 Speaker 1: So okay, did you get instructions on how to write 191 00:11:20,280 --> 00:11:23,479 Speaker 1: effective prompts or did you just kind of do trial. 192 00:11:23,200 --> 00:11:25,439 Speaker 3: And err, no, nothing like that. 193 00:11:25,600 --> 00:11:29,320 Speaker 2: I mean in the artist group itself, there's a lot 194 00:11:29,360 --> 00:11:33,440 Speaker 2: of really amazing and thoughtful creative people who kind of 195 00:11:34,160 --> 00:11:37,040 Speaker 2: show their work and show how they got to make 196 00:11:37,120 --> 00:11:43,480 Speaker 2: the things that they did. But no, not, there was 197 00:11:43,600 --> 00:11:49,480 Speaker 2: no real engineering of our prompts. They were very much 198 00:11:49,720 --> 00:11:55,360 Speaker 2: just play kind of see see what comes out of you. 199 00:11:55,360 --> 00:12:00,040 Speaker 2: You're creative people that we trust, Why don't. 200 00:11:59,880 --> 00:12:03,440 Speaker 3: You just see what works through spaghetti at the wall? 201 00:12:04,360 --> 00:12:07,800 Speaker 1: That's cool. So during the in the piece of mathx 202 00:12:07,880 --> 00:12:11,240 Speaker 1: guide in the interview, some more from shi Kids said 203 00:12:11,240 --> 00:12:14,839 Speaker 1: the Open Eyes researchers they were surprised when they were 204 00:12:14,880 --> 00:12:20,400 Speaker 1: asked about being able to say specific shots. What happened there? 205 00:12:20,840 --> 00:12:23,120 Speaker 1: Was it just that you tried to ask Saora to 206 00:12:23,120 --> 00:12:25,040 Speaker 1: do specific shots and it didn't work, or was it 207 00:12:25,120 --> 00:12:26,040 Speaker 1: just not a feature? 208 00:12:27,760 --> 00:12:30,520 Speaker 2: I think that's maybe taken a little bit out of context. 209 00:12:30,840 --> 00:12:31,599 Speaker 3: I think. 210 00:12:32,880 --> 00:12:38,000 Speaker 2: More so it's just people come from distant, different disciplines. 211 00:12:37,480 --> 00:12:39,079 Speaker 3: And when. 212 00:12:40,760 --> 00:12:43,760 Speaker 2: I say a wide shot on a one hundred and 213 00:12:43,800 --> 00:12:50,160 Speaker 2: thirty millimeter lens, people from my area of expertise know 214 00:12:50,400 --> 00:12:52,360 Speaker 2: sort of immediately what I'm talking about. 215 00:12:52,400 --> 00:12:55,160 Speaker 3: Whereas the researchers, they are. 216 00:12:56,200 --> 00:13:01,440 Speaker 2: More invested in sort of other other things, and so 217 00:13:02,320 --> 00:13:05,839 Speaker 2: it's it's not so much that they didn't understand or 218 00:13:05,920 --> 00:13:08,920 Speaker 2: that sort of didn't understand. It's more so just there's 219 00:13:08,960 --> 00:13:11,280 Speaker 2: all these terms in films. 220 00:13:10,720 --> 00:13:12,400 Speaker 3: Like a zollie or like a. 221 00:13:12,520 --> 00:13:15,800 Speaker 2: Hitchcock zoom or all of these different things that are 222 00:13:16,520 --> 00:13:19,320 Speaker 2: very understandable, but even when you go from set to set, 223 00:13:19,360 --> 00:13:22,680 Speaker 2: they mean something different. So I think it's about trying 224 00:13:22,800 --> 00:13:28,200 Speaker 2: to create a lingua franca between all of these sort 225 00:13:28,240 --> 00:13:34,360 Speaker 2: of different, very different people and very different ways of 226 00:13:34,480 --> 00:13:37,680 Speaker 2: using a tool. What I may call a zoom, you 227 00:13:37,760 --> 00:13:40,360 Speaker 2: may call a dolly shot, et cetera, et cetera. 228 00:13:40,480 --> 00:13:44,559 Speaker 1: So so that feels like a training date, a challenge. 229 00:13:44,760 --> 00:13:49,200 Speaker 2: Yeah, I think it's about trying to figure out how 230 00:13:49,360 --> 00:13:53,520 Speaker 2: and yeah, exactly what to what to train on. 231 00:13:54,600 --> 00:13:58,480 Speaker 1: Yeah, so tell me what was the interface like? Was 232 00:13:58,480 --> 00:14:01,120 Speaker 1: it a chat box? Did you have have? Like? Just 233 00:14:01,160 --> 00:14:02,679 Speaker 1: tell me about what I actually look like. 234 00:14:03,520 --> 00:14:07,839 Speaker 2: Sure, there's limitations of what I can say about things 235 00:14:07,920 --> 00:14:13,480 Speaker 2: like that, but I think the way that I've described 236 00:14:13,480 --> 00:14:16,679 Speaker 2: it to people without giving too much away is I 237 00:14:16,800 --> 00:14:21,040 Speaker 2: think if you're familiar with using something like the Adobe Suite. 238 00:14:21,480 --> 00:14:26,480 Speaker 2: I think that there's some commonalities whether you're using after 239 00:14:26,520 --> 00:14:32,600 Speaker 2: Effects or Premiere or whatever illustrator, there's like commonalities and 240 00:14:32,640 --> 00:14:35,280 Speaker 2: if you can use one, you can sort of flu's 241 00:14:35,320 --> 00:14:39,560 Speaker 2: your way around the others. I would say it's very 242 00:14:39,600 --> 00:14:42,800 Speaker 2: similar like that with open. 243 00:14:42,480 --> 00:14:46,200 Speaker 3: Ayes tools and models that if you are. 244 00:14:47,200 --> 00:14:51,840 Speaker 2: Used to things like chat, GPT and Dolly and those 245 00:14:51,880 --> 00:14:57,360 Speaker 2: types of models, I think you will find it find 246 00:14:57,400 --> 00:14:59,600 Speaker 2: an ease of use in using Zora. 247 00:15:01,400 --> 00:15:04,320 Speaker 1: So within that article they mentioned that there was like 248 00:15:04,320 --> 00:15:07,560 Speaker 1: a three hundred to one shooting ratio, which correct me 249 00:15:07,560 --> 00:15:09,800 Speaker 1: if I'm wrong, means like three hundred seconds of material 250 00:15:10,560 --> 00:15:13,720 Speaker 1: each second of usable material. How does that compare to 251 00:15:14,320 --> 00:15:18,400 Speaker 1: conventional filmmaking in your experience, it. 252 00:15:18,320 --> 00:15:20,920 Speaker 2: Would be even more seconds than that. I would say, 253 00:15:21,160 --> 00:15:26,280 Speaker 2: just three hundred shots at probably ten to twenty seconds apiece. 254 00:15:26,440 --> 00:15:30,080 Speaker 2: So whatever the math is on that, I would say 255 00:15:30,080 --> 00:15:35,000 Speaker 2: that that's pretty common with shooting. You know, when you 256 00:15:35,160 --> 00:15:40,040 Speaker 2: are shooting a fiction film or like even a documentary 257 00:15:40,120 --> 00:15:42,800 Speaker 2: is even crazier for that you shoot all day and 258 00:15:42,840 --> 00:15:47,760 Speaker 2: all day and from We shot a documentary recently and 259 00:15:47,840 --> 00:15:50,280 Speaker 2: I actually had to go back and watch all the dailies, 260 00:15:50,920 --> 00:15:54,560 Speaker 2: we counted about ninety hours of footage that we had, 261 00:15:54,840 --> 00:15:57,920 Speaker 2: and from that nineties hours, you're making an hour and 262 00:15:57,920 --> 00:15:59,800 Speaker 2: a half movie, So you. 263 00:15:59,760 --> 00:16:02,360 Speaker 3: Know, you are really trimming things down. 264 00:16:02,440 --> 00:16:06,600 Speaker 2: And I think also it's like you are getting the 265 00:16:06,720 --> 00:16:11,880 Speaker 2: five seconds that work or the you know, the section 266 00:16:12,200 --> 00:16:15,600 Speaker 2: of that shot that works. And I would say that's 267 00:16:15,600 --> 00:16:17,200 Speaker 2: pretty common to filmmaking. 268 00:16:19,240 --> 00:16:21,920 Speaker 1: How about narrative filmmaking, because I know documentary you have 269 00:16:21,960 --> 00:16:25,120 Speaker 1: a lot of stuff, But I'm just wondering what the 270 00:16:25,160 --> 00:16:28,400 Speaker 1: burden of selection is like compared to the amount of 271 00:16:28,400 --> 00:16:30,760 Speaker 1: shots you take in just a regular movie or regular 272 00:16:30,840 --> 00:16:31,400 Speaker 1: short film. 273 00:16:31,440 --> 00:16:34,160 Speaker 3: Even again, I would. 274 00:16:33,920 --> 00:16:36,520 Speaker 2: Say, at least I can only speak for the way 275 00:16:36,560 --> 00:16:40,160 Speaker 2: that I shoot films. You know, if you had it's subjective. 276 00:16:40,400 --> 00:16:43,560 Speaker 2: It's subjective for sure. If you're David Fincher, you're shooting 277 00:16:43,640 --> 00:16:47,120 Speaker 2: eight hundred takes of like someone picking up a pencil, 278 00:16:47,320 --> 00:16:50,560 Speaker 2: or Stanley Kubrick, you know, is like famous for a 279 00:16:50,680 --> 00:16:55,240 Speaker 2: thousand takes. I would say that the burn rate was 280 00:16:55,320 --> 00:16:59,680 Speaker 2: very similar. I would say that the challenges with Sora 281 00:17:00,480 --> 00:17:05,560 Speaker 2: are like it's unbelievable at making these images that are 282 00:17:06,560 --> 00:17:09,800 Speaker 2: unbelievable and so interesting to look at, But. 283 00:17:11,480 --> 00:17:14,400 Speaker 3: At its current state, it. 284 00:17:14,480 --> 00:17:19,080 Speaker 2: Can sometimes be difficult to do things that in traditional 285 00:17:19,080 --> 00:17:21,880 Speaker 2: shooting would be much easier, where you say, hey, can. 286 00:17:21,680 --> 00:17:23,920 Speaker 3: That guy go over here? 287 00:17:24,040 --> 00:17:26,199 Speaker 2: Or can that person move from one side of the 288 00:17:26,200 --> 00:17:30,600 Speaker 2: screen to the other. Things like that are are more difficult. 289 00:17:30,600 --> 00:17:34,320 Speaker 2: But again this is baby steps. We are in like 290 00:17:34,480 --> 00:17:37,919 Speaker 2: the toddler phase, so I assume that those things will 291 00:17:37,960 --> 00:17:38,400 Speaker 2: get better. 292 00:17:39,880 --> 00:17:44,040 Speaker 1: So you mentioned well shike, it's mentioned in the interview 293 00:17:44,200 --> 00:17:47,080 Speaker 1: the by default it tries to prevent you from creating 294 00:17:47,200 --> 00:17:51,919 Speaker 1: videos that violate copyright law existing copyrights. Did you accidentally 295 00:17:52,840 --> 00:17:55,040 Speaker 1: bump into this regularly or was this something that just 296 00:17:55,080 --> 00:17:56,199 Speaker 1: you didn't really bother you. 297 00:17:57,560 --> 00:18:00,760 Speaker 2: No, you couldn't generate things that So when I was 298 00:18:00,960 --> 00:18:04,960 Speaker 2: mentioning like a Hitchcock zoom, you couldn't mention Hitchcock, So 299 00:18:05,040 --> 00:18:07,480 Speaker 2: you had to find a different way to describe that 300 00:18:07,640 --> 00:18:13,960 Speaker 2: as opposed to like using public figures, anything that would 301 00:18:13,960 --> 00:18:17,119 Speaker 2: have a public figure or a title you would not 302 00:18:17,160 --> 00:18:21,760 Speaker 2: be allowed to generate. From my experience, there wasn't too 303 00:18:21,800 --> 00:18:26,200 Speaker 2: many logos or brands or anything like that, and any 304 00:18:26,240 --> 00:18:28,280 Speaker 2: of the things that I generated, and. 305 00:18:29,600 --> 00:18:32,640 Speaker 1: But something copyright. Did you generate anything that looked copyright? 306 00:18:33,080 --> 00:18:36,680 Speaker 3: No? Not to my not to my eye. 307 00:18:36,760 --> 00:18:41,560 Speaker 1: That's fine. So well, I know you don't know how 308 00:18:41,640 --> 00:18:44,200 Speaker 1: much Sorrow will cost, and we don't know that don't 309 00:18:44,200 --> 00:18:46,920 Speaker 1: even know when it will launch. Can you talk about 310 00:18:46,920 --> 00:18:48,639 Speaker 1: how much you'd be willing to pay for it? What 311 00:18:48,720 --> 00:18:50,600 Speaker 1: do you think it's worth? And I realized that this 312 00:18:50,760 --> 00:18:52,280 Speaker 1: is a vague question. 313 00:18:53,240 --> 00:18:53,760 Speaker 3: For sure. 314 00:18:55,600 --> 00:19:02,840 Speaker 2: I think that there is this illusion that Sora will 315 00:19:02,880 --> 00:19:08,000 Speaker 2: be this solution to all problems, and I don't think 316 00:19:08,040 --> 00:19:10,800 Speaker 2: that that is the case. I think Sora is a 317 00:19:10,840 --> 00:19:15,880 Speaker 2: tool amongst many tools, and for certain things it will 318 00:19:15,920 --> 00:19:16,840 Speaker 2: be very valuable. 319 00:19:17,040 --> 00:19:17,400 Speaker 3: And so. 320 00:19:19,000 --> 00:19:21,280 Speaker 2: In terms of value, it's like, well, how much is 321 00:19:21,320 --> 00:19:24,399 Speaker 2: a glass of water? Well, yes, if a glass of 322 00:19:24,440 --> 00:19:28,080 Speaker 2: water is just like right now in my kitchen, I. 323 00:19:27,560 --> 00:19:29,320 Speaker 3: Wouldn't like to pay that high for it. 324 00:19:29,720 --> 00:19:31,760 Speaker 2: If a glass of water is for a person in 325 00:19:31,800 --> 00:19:34,840 Speaker 2: the desert who desperately needs that glass of water, you 326 00:19:34,920 --> 00:19:37,600 Speaker 2: can really name your price. And I would say that 327 00:19:38,119 --> 00:19:42,240 Speaker 2: for some projects, I think that the usage of Sora 328 00:19:42,400 --> 00:19:44,560 Speaker 2: would be absolutely invaluable, and. 329 00:19:44,560 --> 00:19:47,240 Speaker 3: I would I would. 330 00:19:47,680 --> 00:19:49,680 Speaker 2: I don't know how much exactly that would be, would 331 00:19:49,680 --> 00:19:51,800 Speaker 2: depend on the budget, would depend on the limits and 332 00:19:51,840 --> 00:19:56,640 Speaker 2: the scales, but I would say that there's other projects 333 00:19:56,640 --> 00:19:58,960 Speaker 2: where I think it would be like totally inappropriate or 334 00:19:59,000 --> 00:20:04,600 Speaker 2: like just not worth like what, well, just when I 335 00:20:04,640 --> 00:20:08,280 Speaker 2: think of studio ghibli films that are hand drawn, and 336 00:20:09,760 --> 00:20:12,760 Speaker 2: I think the reason that those films work is because 337 00:20:12,800 --> 00:20:16,080 Speaker 2: of the way that they're made, or I think that 338 00:20:16,119 --> 00:20:19,280 Speaker 2: when you think of art man animation, it's like I 339 00:20:19,320 --> 00:20:21,720 Speaker 2: feel that you could feel the fingerprints in that clay, 340 00:20:22,240 --> 00:20:24,959 Speaker 2: and so I don't think maybe for those types of 341 00:20:25,000 --> 00:20:29,040 Speaker 2: films that it would be appropriate, But I think for 342 00:20:29,119 --> 00:20:31,880 Speaker 2: other types of films like Airhead or others, I think 343 00:20:31,920 --> 00:20:36,960 Speaker 2: it would be extremely appropriate. I think it's up to 344 00:20:37,000 --> 00:20:42,240 Speaker 2: the artists sort of discretion how much they think that 345 00:20:42,240 --> 00:20:43,520 Speaker 2: that tool is needed. 346 00:20:45,000 --> 00:20:50,440 Speaker 1: It's doesn't the inconsistency of shots make this deeply impractical, 347 00:20:50,520 --> 00:20:52,199 Speaker 1: because that's the thing I kept coming back to. 348 00:20:53,000 --> 00:20:55,359 Speaker 2: Yeah, I mean, depends on what project you're working on. 349 00:20:55,400 --> 00:20:58,000 Speaker 2: And again, I think that this is like early days. 350 00:20:58,359 --> 00:21:00,960 Speaker 2: I think that these are kinks and bugs that are 351 00:21:01,119 --> 00:21:07,280 Speaker 2: going to be changed, and already from day one where 352 00:21:07,280 --> 00:21:12,440 Speaker 2: we started using it to where we are today, massive 353 00:21:12,480 --> 00:21:15,919 Speaker 2: improvements have happened, and actually improvements where they've listened to 354 00:21:16,080 --> 00:21:19,919 Speaker 2: things that we have suggested and things that we'd like 355 00:21:20,000 --> 00:21:21,560 Speaker 2: to see and tools we'd. 356 00:21:21,440 --> 00:21:22,040 Speaker 3: Like to see. 357 00:21:22,119 --> 00:21:31,400 Speaker 2: So I think that, for example, for Airhead, the inconsistency 358 00:21:31,520 --> 00:21:38,800 Speaker 2: of having a protagonist, having a protagonist that stays true 359 00:21:39,000 --> 00:21:41,119 Speaker 2: through all these different shots, that's the reason why we 360 00:21:41,160 --> 00:21:43,680 Speaker 2: put a balloon in front of their head, Because while 361 00:21:43,680 --> 00:21:47,760 Speaker 2: different bodies can sort of be accepted, a different face 362 00:21:47,800 --> 00:21:49,400 Speaker 2: and a different head is going to be a little 363 00:21:49,440 --> 00:21:53,880 Speaker 2: bit difficult. And so we turned the limitation into our 364 00:21:54,440 --> 00:21:58,600 Speaker 2: sort of main attribute. And I would say that again, 365 00:21:58,720 --> 00:22:01,719 Speaker 2: that works for that story. But I don't think that 366 00:22:01,840 --> 00:22:06,239 Speaker 2: all stories are going to find this valuable. And I 367 00:22:06,240 --> 00:22:11,280 Speaker 2: also don't think every single shot needs to come from Sora. 368 00:22:11,600 --> 00:22:14,720 Speaker 3: I think that there's a world where it can be. 369 00:22:14,800 --> 00:22:18,399 Speaker 2: An addition, or it can be the start of a 370 00:22:18,480 --> 00:22:21,920 Speaker 2: story where instead of just brainstorming and just having a script, 371 00:22:22,400 --> 00:22:26,119 Speaker 2: you make a sort of moving mood board or a 372 00:22:26,200 --> 00:22:30,320 Speaker 2: trailer or so. I think that there's like tons of 373 00:22:30,400 --> 00:22:35,919 Speaker 2: stages along the pipeline that it would be extremely valuable 374 00:22:36,200 --> 00:22:41,400 Speaker 2: and help elucidate concepts and bring them to life. 375 00:22:41,680 --> 00:22:46,720 Speaker 1: So thematic question, so you avoided filming locations and all 376 00:22:46,720 --> 00:22:49,160 Speaker 1: of this, but you spend a lot of time writing 377 00:22:49,200 --> 00:22:53,360 Speaker 1: prompts and you're waiting for Sora to generate clips, then 378 00:22:53,440 --> 00:22:56,000 Speaker 1: up skating and all that. Do you think you could 379 00:22:56,000 --> 00:22:58,960 Speaker 1: make airhead assuming you could get around the balloon head thing? 380 00:22:59,320 --> 00:23:02,480 Speaker 1: Do you think you could make it quicker in real life? 381 00:23:02,640 --> 00:23:05,000 Speaker 1: Them was soa kind of essential to get it done 382 00:23:05,040 --> 00:23:06,480 Speaker 1: in the timeline you did, because it's like a week 383 00:23:06,520 --> 00:23:07,960 Speaker 1: and a half two weeks, I. 384 00:23:07,920 --> 00:23:13,600 Speaker 2: Think, Yeah, I don't know, that's an interesting question. I mean, 385 00:23:13,600 --> 00:23:15,879 Speaker 2: we definitely wouldn't be able to fly around the world 386 00:23:16,240 --> 00:23:20,560 Speaker 2: and yes, get the shots at the car race and 387 00:23:20,640 --> 00:23:22,000 Speaker 2: all of those things, so. 388 00:23:23,560 --> 00:23:26,199 Speaker 3: I think it would probably be shorter. 389 00:23:26,440 --> 00:23:30,840 Speaker 2: But I think in general, the conversations about like time 390 00:23:30,920 --> 00:23:35,240 Speaker 2: and money are like super reductive in a way in 391 00:23:35,280 --> 00:23:39,760 Speaker 2: that I think that without Sora, this wouldn't exist, And 392 00:23:40,040 --> 00:23:44,160 Speaker 2: I think that that is the more interesting conversation. As 393 00:23:44,880 --> 00:23:48,879 Speaker 2: a director, most directors I know have a folder of 394 00:23:50,359 --> 00:23:53,959 Speaker 2: unrealized ideas, and I think that my hope is that 395 00:23:54,119 --> 00:23:58,160 Speaker 2: Sora will allow us to dust off those folders and 396 00:23:59,359 --> 00:24:02,320 Speaker 2: breathe new life life into concepts, and when people see 397 00:24:02,640 --> 00:24:07,080 Speaker 2: what those concepts could be, my hope is that it 398 00:24:07,640 --> 00:24:13,280 Speaker 2: gives a lot more people opportunities to have their ideas illuminated. 399 00:24:13,720 --> 00:24:16,520 Speaker 2: And whether that means to go and shoot it now 400 00:24:16,560 --> 00:24:20,760 Speaker 2: traditionally or some hybrid. I think that that, to me 401 00:24:20,960 --> 00:24:22,600 Speaker 2: is what's most exciting. 402 00:24:23,960 --> 00:24:26,960 Speaker 1: So where do you see SORA going. I know you're 403 00:24:27,000 --> 00:24:29,400 Speaker 1: considering looking at it as kind of a complementary tool, 404 00:24:30,119 --> 00:24:31,840 Speaker 1: but do you think that that's its use case or 405 00:24:31,880 --> 00:24:34,080 Speaker 1: do you think it'll ever do end to end filmmaking. 406 00:24:35,440 --> 00:24:40,000 Speaker 2: I think I think let a thousand flowers bloom, you know. 407 00:24:40,119 --> 00:24:43,919 Speaker 2: I think that there is people who are going to 408 00:24:44,160 --> 00:24:48,240 Speaker 2: just use it for small complementary things to maybe help 409 00:24:48,320 --> 00:24:50,920 Speaker 2: with in the same way we use stock footage. 410 00:24:50,960 --> 00:24:51,160 Speaker 3: Now. 411 00:24:51,760 --> 00:24:57,560 Speaker 2: I think some people are going to use it as 412 00:24:57,600 --> 00:25:00,679 Speaker 2: a way, say you are from a commune unity that 413 00:25:01,440 --> 00:25:05,720 Speaker 2: has maybe a little bit of a less established film community, 414 00:25:05,760 --> 00:25:09,159 Speaker 2: and it's a way to have you compete with the 415 00:25:09,160 --> 00:25:13,200 Speaker 2: big boys in terms of special effects and usage. And again, 416 00:25:13,280 --> 00:25:16,000 Speaker 2: I don't just think it's as easy as bleep blue 417 00:25:16,000 --> 00:25:19,719 Speaker 2: block type in the prompt here comes the thing, but 418 00:25:19,840 --> 00:25:23,480 Speaker 2: rather it allows you to just have a really powerful 419 00:25:23,640 --> 00:25:28,399 Speaker 2: collaborator that you can help make maybe larger concepts and 420 00:25:28,440 --> 00:25:31,240 Speaker 2: bigger ideas. And then yeah, I think that there's some 421 00:25:31,280 --> 00:25:33,560 Speaker 2: people end to end who are going to make things 422 00:25:33,640 --> 00:25:40,399 Speaker 2: that are completely generated or most of the shots in 423 00:25:40,440 --> 00:25:46,280 Speaker 2: it are generated or things like that. In general, the 424 00:25:46,359 --> 00:25:50,919 Speaker 2: thing that feels interesting to me is like helping to 425 00:25:51,280 --> 00:25:58,199 Speaker 2: deepen humanity, Whereas the more you sort of simplify the process, 426 00:25:58,280 --> 00:26:01,960 Speaker 2: I think that that is like, I don't know, it's 427 00:26:02,040 --> 00:26:07,160 Speaker 2: never a simple process. So anytime you hear about something 428 00:26:07,200 --> 00:26:09,320 Speaker 2: that is going to make it all easy and make 429 00:26:09,359 --> 00:26:11,680 Speaker 2: all your troubles go away, I'd be very wary of that. 430 00:26:11,840 --> 00:26:12,960 Speaker 3: I think film is. 431 00:26:12,880 --> 00:26:17,919 Speaker 2: Going to always be difficult and a challenge, and I 432 00:26:18,040 --> 00:26:24,840 Speaker 2: think the benefit of SORA will be to help lead 433 00:26:24,920 --> 00:26:27,720 Speaker 2: us into new pasts and lead us into new directions. 434 00:26:27,760 --> 00:26:30,240 Speaker 2: If I were to tell you, hey, we made this 435 00:26:30,320 --> 00:26:33,160 Speaker 2: film called Lord of the Rings and it uses CGI 436 00:26:33,400 --> 00:26:37,479 Speaker 2: orcs and it makes massive orc fights. You know, if 437 00:26:37,520 --> 00:26:40,720 Speaker 2: I told you that in the nineteen thirties, you'd probably gasp. 438 00:26:41,800 --> 00:26:44,080 Speaker 2: Or if I told you that CGI is going to 439 00:26:44,080 --> 00:26:46,280 Speaker 2: be a predominant way in which we make films in 440 00:26:46,320 --> 00:26:48,920 Speaker 2: twenty twenty four, I think you would go, ah, that's 441 00:26:48,960 --> 00:26:50,000 Speaker 2: not real filmmaking. 442 00:26:50,640 --> 00:26:53,160 Speaker 1: And I don't think I think you kind of saw 443 00:26:53,160 --> 00:26:54,119 Speaker 1: that in the nineties. 444 00:26:54,320 --> 00:26:59,399 Speaker 2: Really yeah, I don't think history is too kind to 445 00:26:59,440 --> 00:27:03,040 Speaker 2: those people that go, this is not gonna work This 446 00:27:03,160 --> 00:27:06,040 Speaker 2: is not art. This technology is not the way I 447 00:27:06,160 --> 00:27:09,600 Speaker 2: just think it's it depends on the artist, and it 448 00:27:09,600 --> 00:27:11,440 Speaker 2: depends what they want to bring to it. I think 449 00:27:11,480 --> 00:27:14,160 Speaker 2: that's the key X factor here. 450 00:27:15,160 --> 00:27:18,560 Speaker 1: One final question, with that all in mind, do you 451 00:27:18,600 --> 00:27:20,840 Speaker 1: think that SRA is going to hurt filmmakers? Do you 452 00:27:20,880 --> 00:27:22,200 Speaker 1: think it's going to replace people? 453 00:27:23,440 --> 00:27:27,080 Speaker 2: I mean, I hope not. I mean that's my job, 454 00:27:27,320 --> 00:27:29,119 Speaker 2: so I would very hope not. 455 00:27:31,600 --> 00:27:33,919 Speaker 1: No. I very much. 456 00:27:33,840 --> 00:27:42,280 Speaker 2: Understand people's fears, and I think that you know, I'm 457 00:27:42,280 --> 00:27:45,119 Speaker 2: a student of history, so when I look back in 458 00:27:45,320 --> 00:27:52,399 Speaker 2: history and the camera obscura comes out, painters are talking 459 00:27:52,440 --> 00:27:55,159 Speaker 2: about how we aren't going to need painters anymore, because 460 00:27:55,200 --> 00:27:58,919 Speaker 2: now we can capture reality, why do you need a 461 00:27:58,920 --> 00:28:02,000 Speaker 2: painter to go and paint it? And it's a very 462 00:28:02,080 --> 00:28:06,360 Speaker 2: valid point, But painters didn't go away. And then there 463 00:28:06,440 --> 00:28:10,600 Speaker 2: was this whole new industry called photography, and then after photography, 464 00:28:10,600 --> 00:28:13,400 Speaker 2: there was this whole new industry called film. And then 465 00:28:13,440 --> 00:28:16,720 Speaker 2: after film, there was this whole new industry called home video. 466 00:28:17,240 --> 00:28:19,280 Speaker 2: And then after home video, there was this whole new 467 00:28:19,280 --> 00:28:22,760 Speaker 2: industry called cell phone video. And then there was this 468 00:28:22,800 --> 00:28:26,399 Speaker 2: whole new industry called tiktoks and vines, and I just 469 00:28:26,440 --> 00:28:33,320 Speaker 2: think that when people don't come in contacts with things 470 00:28:33,359 --> 00:28:38,280 Speaker 2: they're immediate. As humans, our immediate reaction is fear, and 471 00:28:39,640 --> 00:28:43,960 Speaker 2: we're worried about things that are new because we do 472 00:28:44,040 --> 00:28:48,160 Speaker 2: not yet understand them. And I think that for us, 473 00:28:49,560 --> 00:28:51,920 Speaker 2: we like to face those things face on. And I 474 00:28:52,040 --> 00:28:55,640 Speaker 2: think that the other side of that coin is that 475 00:28:56,000 --> 00:29:00,680 Speaker 2: there's some kid right now in rural Bangladesh who has 476 00:29:00,760 --> 00:29:04,320 Speaker 2: this amazing, big idea and maybe doesn't have all the 477 00:29:04,360 --> 00:29:08,400 Speaker 2: resources that everyone else has, and with these types of technologies, 478 00:29:08,840 --> 00:29:11,719 Speaker 2: it may level the playing field for kids like that 479 00:29:11,960 --> 00:29:15,320 Speaker 2: to compete with the avatars of the world, compete with 480 00:29:15,360 --> 00:29:18,600 Speaker 2: the Marvels of the world, And then I think we're 481 00:29:18,600 --> 00:29:20,640 Speaker 2: going to all be on this level playing field, and 482 00:29:20,680 --> 00:29:23,280 Speaker 2: what's going to matter is not just who has the 483 00:29:23,360 --> 00:29:26,960 Speaker 2: highest budgets and who has the most resources, but who 484 00:29:27,000 --> 00:29:32,080 Speaker 2: has the best stories. And for me, that's the exciting part. 485 00:29:33,840 --> 00:29:37,560 Speaker 2: We work with groups of collaborators that we love and respect, 486 00:29:37,640 --> 00:29:42,320 Speaker 2: and our hope is never let's work with them less. 487 00:29:42,440 --> 00:29:47,160 Speaker 2: Our hope is always let's enrich those relationships and hopefully 488 00:29:47,200 --> 00:29:51,360 Speaker 2: grow them and hopefully bring more people into our collective 489 00:29:51,760 --> 00:29:55,240 Speaker 2: and more people into our process. So that's our hope. 490 00:29:55,560 --> 00:30:00,360 Speaker 2: Maybe I'm utopic, maybe I'm wrong, but that's the that's 491 00:30:00,400 --> 00:30:03,360 Speaker 2: the choice, that that's the way we're choosing to look 492 00:30:03,360 --> 00:30:03,640 Speaker 2: at this. 493 00:30:17,120 --> 00:30:20,840 Speaker 1: In Woodman's mind, Surra is a tool, an extension of 494 00:30:20,880 --> 00:30:25,280 Speaker 1: creatives methods rather than a replacement of filmographers or actors, 495 00:30:25,320 --> 00:30:27,880 Speaker 1: what have you. And that very much lines up with 496 00:30:28,000 --> 00:30:31,560 Speaker 1: sam Ortman an open AI's sales pitch for Sura, his 497 00:30:31,720 --> 00:30:36,080 Speaker 1: utopian perspective, his words, not mine. It's predicated on both 498 00:30:36,120 --> 00:30:39,880 Speaker 1: film studios acting with integrity, something they've proven to never do, 499 00:30:40,320 --> 00:30:43,840 Speaker 1: an open Ai being able to make Sura a significantly 500 00:30:43,920 --> 00:30:47,640 Speaker 1: better tool, something that's going to require masses more training, 501 00:30:47,760 --> 00:30:51,960 Speaker 1: data and compute that I think is actually in existence. 502 00:30:53,240 --> 00:30:56,600 Speaker 1: Paul Trillo, an LA based artist and filmmaker, speaking to 503 00:30:56,680 --> 00:31:00,400 Speaker 1: Business Insider in April, described Saura as a research project 504 00:31:00,400 --> 00:31:03,520 Speaker 1: in Alpha, mentioning that it was a little confusing who 505 00:31:03,520 --> 00:31:06,640 Speaker 1: the market was for the service, and I think that 506 00:31:06,760 --> 00:31:11,120 Speaker 1: jails with another problem that Woodman raised, that what might 507 00:31:11,480 --> 00:31:16,240 Speaker 1: be a zoom out shot for you would be a 508 00:31:16,280 --> 00:31:19,840 Speaker 1: completely different term for someone else, which in turn would 509 00:31:19,920 --> 00:31:22,600 Speaker 1: require open ai to have both the right training data 510 00:31:22,760 --> 00:31:25,040 Speaker 1: of a zoom shot and many, many, many of them, 511 00:31:25,040 --> 00:31:28,880 Speaker 1: to be clear, But they need to know the multitudes 512 00:31:28,920 --> 00:31:33,800 Speaker 1: of different terminologies that go into filmmaking. Now, if they 513 00:31:33,800 --> 00:31:38,000 Speaker 1: don't give a shit, maybe that's a completely different story. 514 00:31:38,080 --> 00:31:42,040 Speaker 1: In short, SAUA faces both the intractable problems of AI 515 00:31:42,120 --> 00:31:44,760 Speaker 1: that I've mentioned in the previous episode, PKI go and 516 00:31:44,760 --> 00:31:47,360 Speaker 1: listen to it, but also a few of its own, 517 00:31:48,280 --> 00:31:52,640 Speaker 1: namely that generating moving images isn't just about ingesting a 518 00:31:52,640 --> 00:31:56,360 Speaker 1: bunch of footage, but it's about understanding said footage well 519 00:31:56,440 --> 00:31:59,600 Speaker 1: enough to generate something else based on a multitude of 520 00:31:59,600 --> 00:32:06,240 Speaker 1: different perspectives, descriptions, and cultural contexts. I'm not sure that 521 00:32:06,360 --> 00:32:10,600 Speaker 1: open AI really Most people realize how complex even the 522 00:32:10,720 --> 00:32:15,840 Speaker 1: simplest movie is, how much work goes into making a film, 523 00:32:16,360 --> 00:32:18,600 Speaker 1: and I think that that's actually what excites people about this, 524 00:32:18,680 --> 00:32:22,080 Speaker 1: because making films can be inefficient, it can be extremely taxing, 525 00:32:22,240 --> 00:32:27,240 Speaker 1: it can be extremely expensive. But the problem here, I'll 526 00:32:27,240 --> 00:32:30,400 Speaker 1: get into the other ones as well, is that SAURA 527 00:32:30,560 --> 00:32:33,720 Speaker 1: is being sold to film studios. That is who Sam 528 00:32:33,760 --> 00:32:37,360 Speaker 1: Mortman is going to, and thus it's going to be 529 00:32:37,400 --> 00:32:39,760 Speaker 1: built for people who don't make movies. I'm actually really 530 00:32:39,840 --> 00:32:42,320 Speaker 1: happy to hear that shy kids and other artists are involved, 531 00:32:42,360 --> 00:32:46,200 Speaker 1: so it'll actually be tuned to be somewhat useful. But 532 00:32:46,280 --> 00:32:50,920 Speaker 1: I don't think people realize how gigantine the task is 533 00:32:50,960 --> 00:32:55,120 Speaker 1: that SRA is going after, and how I think it's 534 00:32:55,160 --> 00:32:59,200 Speaker 1: impossible it can go any further. But I digress. I 535 00:32:59,440 --> 00:33:03,120 Speaker 1: just don't believe that SORA actually works if you're making 536 00:33:03,160 --> 00:33:07,720 Speaker 1: a movie. While pixel movies may take years to render, 537 00:33:07,880 --> 00:33:11,920 Speaker 1: they've got supercomputers and specialized hardware, and more importantly, the 538 00:33:11,960 --> 00:33:14,640 Speaker 1: ability to actually design and move characters in the three 539 00:33:14,720 --> 00:33:19,000 Speaker 1: D space. If you are putting something in Saura, what 540 00:33:19,040 --> 00:33:22,880 Speaker 1: are you designing? If you put a character in this 541 00:33:23,520 --> 00:33:27,640 Speaker 1: in again, you cannot have consistency between these things. That 542 00:33:27,800 --> 00:33:31,440 Speaker 1: is a problem across all generative AI. You can not 543 00:33:31,600 --> 00:33:35,600 Speaker 1: do that unless, of course, using copyrighted footage, mister Oltman. 544 00:33:35,920 --> 00:33:40,720 Speaker 1: But seriously, though, with no consistency cross shots, what the 545 00:33:40,760 --> 00:33:45,000 Speaker 1: hell are you doing? While there are unexpected things that 546 00:33:45,120 --> 00:33:47,560 Speaker 1: might happen in a three D animated movie or a 547 00:33:47,640 --> 00:33:51,600 Speaker 1: CGI situation, you still have complete control over the thing 548 00:33:51,680 --> 00:33:53,720 Speaker 1: you are putting on there, the thing you are animated. 549 00:33:53,760 --> 00:33:56,400 Speaker 1: You can make subtle tweaks to him that doesn't seem 550 00:33:56,400 --> 00:33:59,880 Speaker 1: to be the case with Sora. You can adjust what 551 00:34:00,160 --> 00:34:03,760 Speaker 1: on the screen. But even though this is AI generated, 552 00:34:04,040 --> 00:34:07,960 Speaker 1: it doesn't have the benefits of regular generative stuff like CGI, 553 00:34:08,600 --> 00:34:11,360 Speaker 1: which stands of course for a computer generated image. I believe, 554 00:34:11,400 --> 00:34:12,799 Speaker 1: and if I'm wrong, you're gonna yell at me in 555 00:34:12,800 --> 00:34:17,800 Speaker 1: the emails. But seriously, though the practical use cases for SURA, 556 00:34:18,960 --> 00:34:24,279 Speaker 1: they're just kind of not there. Sora's attempts to replace filmmakers, 557 00:34:24,320 --> 00:34:26,560 Speaker 1: if that is open ayes goal, and I really believe 558 00:34:26,560 --> 00:34:31,200 Speaker 1: it is, they're dead on arrival because it's an impractical 559 00:34:31,200 --> 00:34:34,640 Speaker 1: and ineffective solution and the problems it's solving are really 560 00:34:34,719 --> 00:34:39,360 Speaker 1: only ones created by Hollywood executives. The AI hype bubble, 561 00:34:39,400 --> 00:34:43,320 Speaker 1: as I have noted repeatedly, is one entirely reliant on 562 00:34:43,480 --> 00:34:46,680 Speaker 1: us accepting the idea of what these companies will do, 563 00:34:47,040 --> 00:34:51,400 Speaker 1: rather than interrogating their ability to actually do it. Sourra, 564 00:34:51,680 --> 00:34:55,120 Speaker 1: much like all generative AI, suffers from an imprecision and 565 00:34:55,160 --> 00:34:59,960 Speaker 1: an unreliability caused by hallucinations, an unavoidable result of your 566 00:35:00,080 --> 00:35:04,360 Speaker 1: using mathematics to generate things, and the massive power and 567 00:35:04,400 --> 00:35:08,759 Speaker 1: compute requirements are just prohibitively expensive. If this is going 568 00:35:08,840 --> 00:35:12,279 Speaker 1: to end up as a VFX tool, or a productivity tool, 569 00:35:12,520 --> 00:35:15,920 Speaker 1: or as a fill in tool. It's going to need 570 00:35:16,000 --> 00:35:18,360 Speaker 1: to be a lot cheaper than it is to run. 571 00:35:19,000 --> 00:35:24,319 Speaker 1: Generative AI is already unprofitable to make, soa any kind 572 00:35:24,320 --> 00:35:26,799 Speaker 1: of useful open ay will have to find a way 573 00:35:26,800 --> 00:35:30,600 Speaker 1: to dramatically increase the precision of the prompts, reduce hallucinations 574 00:35:30,600 --> 00:35:34,760 Speaker 1: to pretty much nothing, and vastly increase processing power across 575 00:35:34,800 --> 00:35:38,399 Speaker 1: the board. Sora hasn't even been launched save for, of course, 576 00:35:38,480 --> 00:35:41,520 Speaker 1: these handpicked companies that got to test it, meaning that 577 00:35:41,560 --> 00:35:45,000 Speaker 1: this ten to twenty minute weight between generations of moving 578 00:35:45,000 --> 00:35:48,760 Speaker 1: images that's likely to increase once people use the product. 579 00:35:48,880 --> 00:35:51,799 Speaker 1: And that's before you consider how expensive it's going to 580 00:35:51,840 --> 00:35:54,840 Speaker 1: be to run the bloody thing. This is a significantly 581 00:35:54,920 --> 00:35:59,040 Speaker 1: more complex model than chat GPT, which is already unprofitable. 582 00:36:00,080 --> 00:36:03,080 Speaker 1: Sam Moltman can make money, but can he make profit? 583 00:36:03,760 --> 00:36:07,279 Speaker 1: I severely bloody doubt it. He hasn't before, and I 584 00:36:07,320 --> 00:36:09,840 Speaker 1: don't think he's going to in the future. He's still 585 00:36:10,040 --> 00:36:13,480 Speaker 1: begging Daddy Satchia over at Microsoft to give him a 586 00:36:13,520 --> 00:36:17,360 Speaker 1: supercomputer so his things can fart out things more profitably. 587 00:36:17,400 --> 00:36:22,319 Speaker 1: It's just drives me a little insane. And these things 588 00:36:22,360 --> 00:36:25,680 Speaker 1: I've talked about their intractable problems that open aiy has 589 00:36:25,680 --> 00:36:29,040 Speaker 1: failed to solve. They've failed to make a more efficient 590 00:36:29,080 --> 00:36:31,359 Speaker 1: model for Microsoft last year in twenty twenty three, their 591 00:36:31,440 --> 00:36:35,879 Speaker 1: Arakis model Jesus Christ. And while GPT five is meant 592 00:36:35,920 --> 00:36:38,919 Speaker 1: to be materially better, to quote mister Altman, it isn't 593 00:36:38,960 --> 00:36:42,880 Speaker 1: obvious what better means when GPT four performs worse at 594 00:36:42,880 --> 00:36:46,840 Speaker 1: some tasks than its predecessor. I do believe Sam Mortman 595 00:36:46,920 --> 00:36:48,600 Speaker 1: is telling the truth when he says that the future 596 00:36:48,640 --> 00:36:51,799 Speaker 1: of AI requires an energy breakthrough. But the thing I 597 00:36:51,800 --> 00:36:54,400 Speaker 1: think he's leaving out is that it may take an 598 00:36:54,480 --> 00:36:58,440 Speaker 1: energy breakthrough and indeed more chips for generative AI to 599 00:36:58,560 --> 00:37:03,160 Speaker 1: approach any level of ness. And he's hoping that people 600 00:37:03,200 --> 00:37:06,280 Speaker 1: will buy the hype without asking too many annoying questions 601 00:37:06,280 --> 00:37:09,600 Speaker 1: like what does this stuff actually do? Or is this useful? 602 00:37:09,840 --> 00:37:12,520 Speaker 1: Or does this actually help me? Or will this be 603 00:37:12,640 --> 00:37:16,160 Speaker 1: around in ten years? To be clear, Sam Altman is 604 00:37:16,200 --> 00:37:19,520 Speaker 1: the single most well connected and well funded man in AI, 605 00:37:19,880 --> 00:37:23,600 Speaker 1: with a direct connection to Microsoft, a multi trillion dollar 606 00:37:23,640 --> 00:37:27,760 Speaker 1: tech company, and a rollodexter includes effectively every major founder 607 00:37:27,800 --> 00:37:31,000 Speaker 1: of the last decade, and he still can't get past 608 00:37:31,160 --> 00:37:34,440 Speaker 1: any of these problems, partly because he is not technical 609 00:37:34,600 --> 00:37:37,200 Speaker 1: and thus can't really solve the problems himself, and partly 610 00:37:37,200 --> 00:37:40,320 Speaker 1: because the problems he's facing are burdened by the laws 611 00:37:40,360 --> 00:37:45,720 Speaker 1: of maths and physics. Generative AI hallucinates because it doesn't 612 00:37:45,760 --> 00:37:49,000 Speaker 1: have a consciousness or any ability to learn or know anything. 613 00:37:50,080 --> 00:37:54,480 Speaker 1: It's extremely expensive because even the simplest prompts require GPT 614 00:37:54,560 --> 00:37:59,000 Speaker 1: four to run highly complex mathematical equations on graphics processing 615 00:37:59,120 --> 00:38:03,240 Speaker 1: units that cost upwards of ten thousand dollars apiece. Even 616 00:38:03,320 --> 00:38:06,520 Speaker 1: if generative AI were cheaper or more efficient or required 617 00:38:06,600 --> 00:38:10,200 Speaker 1: less power, it would still be a process that generates 618 00:38:10,239 --> 00:38:13,319 Speaker 1: answers based on the extremely complex process of ingesting an 619 00:38:13,360 --> 00:38:18,120 Speaker 1: increasingly dwindling amount of training data. These problems are significantly 620 00:38:18,160 --> 00:38:22,160 Speaker 1: compounded when you consider the complexity, size, and massive legal 621 00:38:22,239 --> 00:38:26,600 Speaker 1: ramifications of training a model on videos. A problem that 622 00:38:26,640 --> 00:38:30,279 Speaker 1: nobody has seem fit to push Altmnormorti or anyone else 623 00:38:30,280 --> 00:38:34,279 Speaker 1: at Open AI about what's a pisstake really seems like 624 00:38:34,280 --> 00:38:37,640 Speaker 1: an obvious one, like, hey man, you need a bunch 625 00:38:37,719 --> 00:38:41,200 Speaker 1: of training data to train chat GPT, which does words 626 00:38:41,360 --> 00:38:43,880 Speaker 1: how are you getting all these videos again? Big credit 627 00:38:43,920 --> 00:38:47,839 Speaker 1: to Joanna Stern who asked mirror Murati, CTO of open Ai, 628 00:38:47,960 --> 00:38:52,080 Speaker 1: whether Sawer was trained on YouTube videos, and then Mirrormorati 629 00:38:52,120 --> 00:38:55,440 Speaker 1: of course made that incredible face. Go look up that video. 630 00:38:55,560 --> 00:38:58,840 Speaker 1: I'll link it in the notes. That's how moately the 631 00:38:58,920 --> 00:39:02,640 Speaker 1: problem with the current bubble. So much of its success 632 00:39:02,680 --> 00:39:05,760 Speaker 1: requires us to tolerate and applaud these half fast, half 633 00:39:05,800 --> 00:39:08,880 Speaker 1: finished tools that only sort of kind of do the 634 00:39:08,920 --> 00:39:10,880 Speaker 1: things they're meant to do, and we're meant to nod 635 00:39:10,960 --> 00:39:14,279 Speaker 1: and smile and clap and say great job, Sammy, like 636 00:39:14,320 --> 00:39:17,240 Speaker 1: we're talking to a bloody child rather than a startup 637 00:39:17,280 --> 00:39:20,200 Speaker 1: with thirteen billion dollars in funding with a CEO that 638 00:39:20,280 --> 00:39:24,080 Speaker 1: has the backing of goddamn Microsoft and soa is the 639 00:39:24,160 --> 00:39:29,120 Speaker 1: ugliest messiest problem of them all. It's videos, while superficially impressive, 640 00:39:29,280 --> 00:39:32,439 Speaker 1: are still deeply, deeply flawed. They take way too long 641 00:39:32,440 --> 00:39:34,560 Speaker 1: to generate a problem that's only going to get worse, 642 00:39:35,040 --> 00:39:38,320 Speaker 1: and they're just far too inconsistent, which is a problem 643 00:39:38,360 --> 00:39:42,000 Speaker 1: created by the nature of how generative AI works and 644 00:39:42,040 --> 00:39:47,759 Speaker 1: its approach to generating things using mathematics, and if it's 645 00:39:47,800 --> 00:39:49,960 Speaker 1: planning to be a VFX tool, if it's planning to 646 00:39:49,960 --> 00:39:54,040 Speaker 1: be a sidearm for filmographers, it's going to have to 647 00:39:54,080 --> 00:39:59,400 Speaker 1: be a lot cheaper than it's really practical to make it. Again, 648 00:40:00,160 --> 00:40:03,879 Speaker 1: nothing open Ai makes is profitable. They may make over 649 00:40:03,920 --> 00:40:07,440 Speaker 1: a billion dollars of revenue, but everything is burning money. 650 00:40:08,480 --> 00:40:14,600 Speaker 1: It's just very frustrating. It's all very frustrating. Sora seems 651 00:40:14,719 --> 00:40:17,920 Speaker 1: kind of cool, but when you take away the cool 652 00:40:18,400 --> 00:40:20,440 Speaker 1: side and you just look at it for what it is, 653 00:40:20,920 --> 00:40:23,800 Speaker 1: it's just another con from Sam Altman. It's just another 654 00:40:24,120 --> 00:40:27,480 Speaker 1: unfinished product that is not able to fit the task. 655 00:40:28,440 --> 00:40:31,040 Speaker 1: It's just another thing that you look at and you say, oh, 656 00:40:31,120 --> 00:40:33,120 Speaker 1: if that was just a bit better, it'd be really good. 657 00:40:33,200 --> 00:40:36,560 Speaker 1: Except in this case it would be a lot better. Yeah, 658 00:40:36,840 --> 00:40:39,839 Speaker 1: all the press writes about it's incredible, it's amazing, and 659 00:40:40,480 --> 00:40:44,520 Speaker 1: you can separate the technological achievement of using maths to 660 00:40:44,560 --> 00:40:50,120 Speaker 1: generate a visual moving image that's genuinely cool. But you 661 00:40:50,280 --> 00:40:52,759 Speaker 1: gotta stop for a second and say, as cool as 662 00:40:52,800 --> 00:40:55,399 Speaker 1: this is, the people in the back of their shot, 663 00:40:55,440 --> 00:41:00,279 Speaker 1: they're molding into each other. It's like the thing, it's disgusting. Hey, 664 00:41:00,360 --> 00:41:04,640 Speaker 1: that monkey's got like five arms. That's weird. I don't know. 665 00:41:04,800 --> 00:41:09,160 Speaker 1: I just feel like normal people don't get this much leniency. 666 00:41:10,080 --> 00:41:12,920 Speaker 1: You and I don't get people saying great job when 667 00:41:12,960 --> 00:41:15,640 Speaker 1: we do kind of a shitty job. And if we 668 00:41:15,840 --> 00:41:19,920 Speaker 1: brought something to someone that was insanely expensive only really 669 00:41:19,920 --> 00:41:22,000 Speaker 1: did ten percent of the job, you needed it too, 670 00:41:22,800 --> 00:41:26,680 Speaker 1: And also the things that created took forever and looked horrifying. 671 00:41:27,080 --> 00:41:29,799 Speaker 1: I don't think we'd get told great job. I think 672 00:41:29,840 --> 00:41:32,400 Speaker 1: we'd be told we'd wasted a lot of money and 673 00:41:32,440 --> 00:41:36,279 Speaker 1: that someone was quite mad at us. I'm tired of this. 674 00:41:36,960 --> 00:41:41,040 Speaker 1: I'm tired of these companies announcing these half completed products 675 00:41:41,200 --> 00:41:43,319 Speaker 1: and having the media dance around and act like they've 676 00:41:43,320 --> 00:41:47,439 Speaker 1: delivered something truly incredible. I'm tired of the public being 677 00:41:47,520 --> 00:41:50,759 Speaker 1: expected to do the mental and emotional labor for Sam 678 00:41:50,800 --> 00:41:53,960 Speaker 1: Moultman and other AI companies, saying it's remarkable that they're 679 00:41:54,000 --> 00:41:56,919 Speaker 1: even able to do this, and assume and give them 680 00:41:56,960 --> 00:42:00,760 Speaker 1: credit for some inevitable future where all of thesebms are gone, 681 00:42:00,880 --> 00:42:03,520 Speaker 1: despite little proof that such a thing is possible and 682 00:42:03,560 --> 00:42:08,080 Speaker 1: plenty of proof that it isn't. And as I've suggested, 683 00:42:08,560 --> 00:42:11,319 Speaker 1: I really don't think it is. I think Sora is 684 00:42:11,360 --> 00:42:15,799 Speaker 1: dead on arrival. I think it's too expensive, too imprecise, 685 00:42:15,840 --> 00:42:19,400 Speaker 1: and there is no fixing those problems. You can iterate 686 00:42:19,440 --> 00:42:22,120 Speaker 1: on them, you can improve them, but without some kind 687 00:42:22,120 --> 00:42:24,799 Speaker 1: of energy or chips breakthrough, they're not even going to 688 00:42:24,840 --> 00:42:28,640 Speaker 1: have the compute or really the money to build this 689 00:42:28,680 --> 00:42:32,440 Speaker 1: thing into anything even half functional. And I'm calling on 690 00:42:32,480 --> 00:42:35,440 Speaker 1: the press to push back on these companies. I'm calling 691 00:42:35,440 --> 00:42:39,719 Speaker 1: on them to refuse to declare this quasi functional software 692 00:42:40,239 --> 00:42:46,080 Speaker 1: as complete. I'm tired of seeing the media back these 693 00:42:46,120 --> 00:42:50,320 Speaker 1: companies and do marketing work for them when they're not done. 694 00:42:50,719 --> 00:42:55,120 Speaker 1: They don't deserve the credit, and I'm demanding that people 695 00:42:55,160 --> 00:42:59,520 Speaker 1: like Sam Altman actually change the world before anyone says 696 00:42:59,520 --> 00:43:00,200 Speaker 1: that they're doing. 697 00:43:00,280 --> 00:43:00,320 Speaker 3: So. 698 00:43:08,600 --> 00:43:11,040 Speaker 1: Thank you for listening to Better Offline. The editor and 699 00:43:11,040 --> 00:43:14,239 Speaker 1: composer of the Better Offline theme song is Matasowski. You 700 00:43:14,239 --> 00:43:16,480 Speaker 1: can check out more of his music and audio projects 701 00:43:16,640 --> 00:43:20,160 Speaker 1: at Mattasowski dot com M A T T O. S 702 00:43:20,200 --> 00:43:24,279 Speaker 1: O W s KI dot com. You can email me 703 00:43:24,320 --> 00:43:26,919 Speaker 1: at easy at Better Offline dot com, or visit Better 704 00:43:26,960 --> 00:43:29,399 Speaker 1: Offline dot com to find more podcast links and of course, 705 00:43:29,440 --> 00:43:32,560 Speaker 1: my newsletter. I also really recommend you go to chat 706 00:43:32,600 --> 00:43:35,239 Speaker 1: dot Where's youreed dot at to visit the discord, and 707 00:43:35,280 --> 00:43:37,960 Speaker 1: go to our slash Better Offline to check out our reddit. 708 00:43:38,760 --> 00:43:42,000 Speaker 1: Thank you so much for listening. Better Offline is a 709 00:43:42,040 --> 00:43:45,120 Speaker 1: production of cool Zone Media. For more from cool Zone Media, 710 00:43:45,239 --> 00:43:48,400 Speaker 1: visit our website cool Zonemedia dot com, or check us 711 00:43:48,400 --> 00:43:51,360 Speaker 1: out on the iHeartRadio app, Apple Podcasts, or wherever you 712 00:43:51,440 --> 00:44:12,800 Speaker 1: get your podcasts.