WEBVTT - Radio Better Offline: David "Shingy" Shing 0:00:02.400 --> 0:00:08.160 Also media, Hi, Mimo, moo ed zitron. This is better offline. 0:00:08.160 --> 0:00:23.440 We're in iHeart Radio studio in beautiful New York City today. 0:00:23.520 --> 0:00:25.599 I am joined by a man who in twenty fourteen 0:00:25.680 --> 0:00:28.680 roast to fame after becoming AOL's digital profit. He joins 0:00:28.720 --> 0:00:32.760 me in the studio today, David Shing, best known as Shingy. David, Hey, ed, 0:00:32.800 --> 0:00:35.040 how are you, mate, sing? I am fantastic an intro. 0:00:35.320 --> 0:00:37.840 Oh yeah, this is the most normal show in tech. 0:00:38.680 --> 0:00:40.280 What are you up to these days? Shingy? 0:00:41.360 --> 0:00:44.920 Great question. There's three things. I'm primarily up to them. Okay, 0:00:45.040 --> 0:00:47.680 I'm out speaking and educating, that's right, what I'm known for. 0:00:47.760 --> 0:00:50.120 I also have a creative house, nice that started around 0:00:50.240 --> 0:00:51.200 pandemic time. 0:00:51.760 --> 0:00:52.800 What's the creative house too? 0:00:53.080 --> 0:00:56.200 It does everything from iconography all the way through the strategy, 0:00:56.240 --> 0:00:59.080 so all the way through psychography. You happened to be 0:00:59.080 --> 0:01:01.440 wearing a brand Onion T shirt. Yeah, somebody designed that, dude. 0:01:01.480 --> 0:01:03.080 Okay, so a design thing cool. 0:01:03.120 --> 0:01:06.040 And then the other flip side is an advisory practice. 0:01:06.120 --> 0:01:09.600 That's cool, and it's institutional advisory as well as small startups, 0:01:09.600 --> 0:01:11.800 which has really been research for me, which has been amazing. 0:01:12.000 --> 0:01:15.080 Yeah, as a fellow consultant with the podcast. It's always 0:01:15.080 --> 0:01:18.240 good to meet another one. So you it's been a 0:01:18.240 --> 0:01:20.880 while since you rose to fame. What kind of happened? 0:01:20.880 --> 0:01:23.200 What kind of happened? Because you I remember I was 0:01:23.240 --> 0:01:24.840 in the Bay Area at the time when I saw 0:01:24.880 --> 0:01:29.440 you you arise on MSNBC, and then just went and 0:01:29.480 --> 0:01:31.480 did other things. I had to go and make some 0:01:31.520 --> 0:01:34.760 mistakes in my life. I guess what was it you 0:01:34.800 --> 0:01:35.240 did today? 0:01:35.440 --> 0:01:37.680 Well, so I ran thirteen countries for them. I ran 0:01:37.720 --> 0:01:40.800 the media in marketing for them right throughout Europe. 0:01:41.400 --> 0:01:43.200 But what did that actually entail? 0:01:43.560 --> 0:01:46.240 I ran thirteen countries with multi millions? 0:01:46.280 --> 0:01:47.520 But what did you do each day? 0:01:47.600 --> 0:01:50.720 Because I launched I launched a series of websites, and 0:01:50.760 --> 0:01:54.640 I launched a series of ad platforms, and I helped 0:01:54.680 --> 0:01:56.800 change the iconography for AOL. And when you build a 0:01:56.800 --> 0:01:59.520 new brand and you don't have any budget to support it, right, 0:01:59.720 --> 0:02:01.600 you have to think about a way to be in 0:02:01.600 --> 0:02:02.240 the marketplace. 0:02:02.280 --> 0:02:04.680 So I created a PERSONA wait, did you not have 0:02:05.240 --> 0:02:06.080 like much budget? 0:02:06.320 --> 0:02:06.480 Oh? 0:02:06.520 --> 0:02:09.240 Yeah, we did, but not in Europe. No, not in Europe. 0:02:09.400 --> 0:02:12.239 No, Because what happened in a big company, you could 0:02:12.240 --> 0:02:17.480 probably appreciate this if one division performs poorly, everyone suffers, right, 0:02:17.639 --> 0:02:19.359 So if you're not close to the sun, you're not 0:02:19.360 --> 0:02:20.480 going to get all the budget. 0:02:20.800 --> 0:02:23.560 It's just how it works. It's funny because your whole thing, 0:02:23.600 --> 0:02:25.400 and I don't say this is an ensult, got kind 0:02:25.440 --> 0:02:28.200 of flattenedto that one picture of you with the kind 0:02:28.200 --> 0:02:33.000 of yeah because and then there was that Guardian interview. 0:02:33.080 --> 0:02:36.240 It's like you drew a zebra. It said you showed 0:02:36.240 --> 0:02:39.800 someone AOL a picture of a zebra zebra pant pants. 0:02:39.919 --> 0:02:42.840 Possibly it's just I was wearing them. You're wearing a 0:02:42.919 --> 0:02:48.959 nice right. Yeah. It's it's funny because you've become somewhat 0:02:49.000 --> 0:02:52.240 of like a like a I want to say, like 0:02:52.240 --> 0:02:55.680 a character within meme, a meme. I think memes fair, 0:02:55.720 --> 0:02:56.320 but it's more. 0:02:56.240 --> 0:02:58.160 Than character's probably a little underrated. 0:02:58.360 --> 0:03:02.200 Yeah, it's just because you have a real job, it seems, 0:03:02.240 --> 0:03:04.200 and you have this whole time, despite the fact that 0:03:04.240 --> 0:03:06.959 I've been told, like, oh, this guy just like shows 0:03:07.040 --> 0:03:08.880 up and says he's the digital profit and I was 0:03:08.960 --> 0:03:10.880 genuinely curious about like, I. 0:03:11.800 --> 0:03:14.360 Never really told anybody the mystery. I want to hear 0:03:14.360 --> 0:03:18.040 that and the magic of it still replicates today. So 0:03:18.560 --> 0:03:23.040 I would go in and get interviews, get audiences with 0:03:23.120 --> 0:03:24.880 brands we could never be in front. 0:03:24.639 --> 0:03:27.880 Of like what like what brand? Like? Yeah, what kind 0:03:27.880 --> 0:03:30.760 of brands? Nike? Nike? Why couldn't al get in front 0:03:30.760 --> 0:03:30.960 of them? 0:03:31.000 --> 0:03:32.919 Well, they could, but they're going to just sell ads. 0:03:32.960 --> 0:03:34.880 I'm trying to sell innovation on top of the ads, 0:03:35.040 --> 0:03:37.360 so much bigger than we would ever be able to do. 0:03:37.400 --> 0:03:39.320 And what would that innovation manifest? That's what were you 0:03:39.360 --> 0:03:41.440 selling to Nike? For example? I know this is just 0:03:41.480 --> 0:03:42.119 an example. 0:03:43.040 --> 0:03:45.960 We invented an ad called Devil, which was like this 0:03:46.080 --> 0:03:52.080 incredible new magazine esque style ad takeover right, that the 0:03:52.200 --> 0:03:54.120 IAB ended up picking up and running with. So we 0:03:54.160 --> 0:03:56.960 were just building things that were radically different than spots 0:03:56.960 --> 0:03:58.320 and dots that sales teams are selling. 0:03:58.520 --> 0:04:01.880 Right, So how do you feel about AI? How are 0:04:01.880 --> 0:04:04.720 you feeling about all this? Because look, you were part 0:04:04.800 --> 0:04:07.840 of a hype cycle. So end of twenty fourteen is 0:04:07.880 --> 0:04:10.640 like it was like a very hype driven time. It's 0:04:10.680 --> 0:04:14.400 like the sexiest time in indiego Go and Kickstarter land. 0:04:14.400 --> 0:04:17.799 I would say, back then, what do you think of AI? 0:04:17.839 --> 0:04:19.160 How do you feel about it? Do you think it's 0:04:19.160 --> 0:04:21.640 a bubble? No? 0:04:21.640 --> 0:04:24.400 No, Well, it's also been around for forty fifty. 0:04:24.240 --> 0:04:27.320 Years, right, so generative AI specifically. 0:04:27.240 --> 0:04:28.120 What do I think about that? 0:04:28.200 --> 0:04:28.400 Yeah? 0:04:28.400 --> 0:04:32.000 Sure, think it's magical. Really Yeah, it hallucinates occasionally, but 0:04:32.040 --> 0:04:35.080 I think it's great because if I can extend a 0:04:35.120 --> 0:04:37.320 background without having to go reshoot it, that's pretty good 0:04:37.400 --> 0:04:38.960 and I'm able to change it out. 0:04:38.560 --> 0:04:39.680 Can't do that though. 0:04:39.839 --> 0:04:41.800 Yeah, you're not been you know, haven't checked out fire 0:04:41.839 --> 0:04:42.320 Fly or. 0:04:42.480 --> 0:04:45.480 I have I read your sub stack. I do read 0:04:45.520 --> 0:04:49.279 your blog. It's just right now as a hype man. 0:04:49.400 --> 0:04:52.080 I don't mean that derisively. I mean that, like your 0:04:52.960 --> 0:04:56.760 job is ostensibly hype. Right now, it feels like there 0:04:56.839 --> 0:05:02.440 is this marketing dissonance between in Firefly, which I'm aware 0:05:02.520 --> 0:05:04.120 and I know there are lots of people who listen 0:05:04.120 --> 0:05:05.680 to the showho are going to be very angry at 0:05:05.680 --> 0:05:08.440 me that we're even talking about Adobe AI. Calm down, everyone, 0:05:08.480 --> 0:05:12.279 it's shingy. Allow him in. But it's it feels like 0:05:12.320 --> 0:05:15.680 you've got this massive business failure happening in the background, 0:05:15.680 --> 0:05:18.040 billions of dollars burned from Open AI, but you've got 0:05:18.320 --> 0:05:20.600 some utility. How do you balance that? 0:05:20.680 --> 0:05:24.480 Like, how do you in what characterizational you said? 0:05:24.520 --> 0:05:28.200 Generally they so, well, I'm talking about the fact that 0:05:28.880 --> 0:05:32.400 you're hyping up something that is unsustainable right now, which 0:05:32.440 --> 0:05:36.719 bit I mean the generative AI features of Firefly, for example, 0:05:36.760 --> 0:05:37.400 the same. 0:05:37.200 --> 0:05:38.120 Is it not sustainable? 0:05:38.279 --> 0:05:40.480 Well, because open AI burns five billion dollars a year, 0:05:40.520 --> 0:05:42.599 they still haven't worked out any profitability for any of 0:05:42.640 --> 0:05:47.800 these models. Yes, but fireflies run on generitive models. 0:05:47.560 --> 0:05:53.200 Yeah, right, and they help create ad performances sure at scale. Right, 0:05:53.560 --> 0:05:55.920 So at some point in time these things level. 0:05:55.720 --> 0:05:58.600 Out when because that's the thing, like the amount when 0:05:58.600 --> 0:06:03.400 there's demand for better right, I just feel like we're 0:06:03.440 --> 0:06:06.600 a little lost within the general of AI conversation right 0:06:06.640 --> 0:06:09.919 as an industry. Right, And I'm curious you see, do 0:06:09.960 --> 0:06:13.560 you play with Runway. I've played with Sora a bit. Yeah, 0:06:13.560 --> 0:06:16.159 I just think I found Runway to be really fucking mediocre. 0:06:16.480 --> 0:06:19.680 Well, there's a dystopian to it, which is definitely that 0:06:19.720 --> 0:06:21.360 shine will come off and that production is going to 0:06:21.360 --> 0:06:24.840 be pretty amazing. Single single tool that does one thing 0:06:25.120 --> 0:06:27.120 no good, right, but a tool that allows you to 0:06:27.160 --> 0:06:29.839 master audio without an engineer pretty amazing. 0:06:30.000 --> 0:06:31.520 Yeah, But does that exist yet? 0:06:31.680 --> 0:06:31.840 Yeah? 0:06:31.839 --> 0:06:35.440 It does. With voice Riverside is not good. 0:06:35.600 --> 0:06:37.640 Riverside is a podcasting platform. 0:06:37.200 --> 0:06:38.880 And they have AI mastering on it. 0:06:38.880 --> 0:06:41.320 God bless them, but they're not just doing mastering. So 0:06:41.360 --> 0:06:43.479 you can upload something to something like voice AI and 0:06:43.520 --> 0:06:45.600 it's absolutely designed for mastering, right. 0:06:45.640 --> 0:06:50.280 But there is this massive financial problem at the side 0:06:50.279 --> 0:06:52.480 of this, that this stuff is burning so much money. 0:06:52.480 --> 0:06:55.200 And I guess you could have on device models, but 0:06:55.480 --> 0:06:59.160 that's nowhere near what I just I don't know how 0:06:59.200 --> 0:07:01.080 you couch these two things. 0:07:01.720 --> 0:07:04.800 Supply and demand. I mean, there's a massive supply for 0:07:05.520 --> 0:07:10.160 new ways of creating different tools, and the supply cycle 0:07:10.240 --> 0:07:12.960 is incredibly wide. It feels like, well, you know, back 0:07:13.000 --> 0:07:16.600 in the web one dot ohs cycles, right, but now 0:07:16.600 --> 0:07:18.400 it has different tools that are much faster build. 0:07:18.560 --> 0:07:22.720 But the demand isn't there I think. I mean, I 0:07:22.760 --> 0:07:24.920 did a piece a few months ago and weeks ago 0:07:25.040 --> 0:07:28.000 Jesus Christ time dilation where it was like co pilot 0:07:28.160 --> 0:07:31.920 has like eleven million monthly active users, okay, which is 0:07:32.160 --> 0:07:35.800 pretty pispor and that's Microsoft's I am hearing that there 0:07:35.800 --> 0:07:37.800 are various companies like eleven Labs which are kind of 0:07:37.880 --> 0:07:40.720 leveling out as far as user base goes. What happens 0:07:40.760 --> 0:07:42.239 if this doesn't get much bigger? 0:07:42.320 --> 0:07:44.000 It just continues to do what everything else does. It 0:07:44.000 --> 0:07:47.360 tails off and becomes a niche, which is okay, you think, so, yeah, 0:07:47.360 --> 0:07:48.200 it's totally okay. 0:07:48.400 --> 0:07:49.880 So you seem quite eye on it. So what are 0:07:49.880 --> 0:07:52.360 you using? Then? Talk to me about the tools you use? 0:07:53.280 --> 0:07:53.840 The obvious. 0:07:54.040 --> 0:07:57.600 I mean, I think chat's interesting. I think that Claude 0:07:57.640 --> 0:07:59.920 is interesting to help you do draft one. So those 0:08:00.320 --> 0:08:04.239 tools are fine. I think sora is it seems interesting. 0:08:04.280 --> 0:08:07.440 But Dystopian Runaway works okay if you can actually model 0:08:07.440 --> 0:08:07.960 it correctly. 0:08:08.800 --> 0:08:11.840 When you say dystopian, what do you mean over glossy? 0:08:12.240 --> 0:08:16.680 Really hard to understand. Depth of field like it's if 0:08:16.720 --> 0:08:19.320 you are going to be a deep practitioner, you know, 0:08:19.680 --> 0:08:22.480 the sort of thing you want to try and visualize, 0:08:22.480 --> 0:08:24.640 that can be very difficult. And I think that's kind 0:08:24.640 --> 0:08:28.360 of the challenge today is these tools do remarkable. 0:08:27.880 --> 0:08:28.600 Amount of work. 0:08:29.160 --> 0:08:31.440 It's a matter of can you get it to do 0:08:32.320 --> 0:08:34.880 the work that you wanted to do without taking more 0:08:34.920 --> 0:08:38.600 time than actually doing it physically. It's software, So it's 0:08:38.600 --> 0:08:43.320 just software, and it's and it's curious, but it's not it. 0:08:43.520 --> 0:08:45.839 Here's what I'm am saying. I'm seeing a consolidation of 0:08:45.880 --> 0:08:47.760 tools saying here's one tool that you log into and 0:08:47.760 --> 0:08:49.640 it does all these things for you, but it does 0:08:49.679 --> 0:08:52.320 all of them only, Okay, so you have to splinter 0:08:52.360 --> 0:08:54.199 off and find something that does just what you want 0:08:54.240 --> 0:08:54.520 it to do. 0:08:54.679 --> 0:08:56.480 So focused, and you know. 0:08:56.520 --> 0:08:59.240 Better at book writing then maybe stories, or better that 0:08:59.760 --> 0:09:03.640 draw oring than maybe paining a landscape. I mean these 0:09:03.840 --> 0:09:08.239 articulations that become really just splinters of a certain technique 0:09:08.679 --> 0:09:11.120 today or crud into one thing. And I don't think 0:09:11.280 --> 0:09:13.600 anything does a great job. Words are great. 0:09:14.080 --> 0:09:16.800 I think the agent if I agree on the words side, 0:09:16.920 --> 0:09:20.080 I don't think well as at and you're still your 0:09:20.320 --> 0:09:22.880 points made. But it's way better than me as a 0:09:22.880 --> 0:09:26.480 first edit copywriter. So yeah, but that's a skill issue, shingy, 0:09:26.840 --> 0:09:29.360 like you could get better by writing more. You could. 0:09:29.640 --> 0:09:32.880 It's good, Like do you is your substack chat GPT 0:09:32.960 --> 0:09:36.000 written or claud written? No, but it's but it's you know. 0:09:36.880 --> 0:09:39.719 I think what's interesting about the tools I don't want 0:09:39.720 --> 0:09:42.280 to have to necessarily become very technical when I'm wanting 0:09:42.320 --> 0:09:45.040 to shoot an idea of a landscape or something. By 0:09:45.080 --> 0:09:47.760 the way, I'm only using my own images and only 0:09:48.760 --> 0:09:52.600 purposely not using anything generative. It's all like photos from 0:09:52.600 --> 0:09:56.880 my own photo library, right, no stock, no, no generative 0:09:56.880 --> 0:09:59.560 because until it's actually really good. As much as I'm 0:09:59.559 --> 0:10:02.240 an advocate for it, I'm a better believer in craft 0:10:02.240 --> 0:10:05.160 than I am. Then so can these tools like mastering 0:10:05.200 --> 0:10:07.280 for example of sound love that love that for a 0:10:07.320 --> 0:10:09.760 tool because I have no clue how to bring eqs up. 0:10:09.760 --> 0:10:12.079 I have no clue how to bring in any depth 0:10:12.120 --> 0:10:13.679 within a voice, and no clue on that. 0:10:13.720 --> 0:10:16.480 So God bless, and do you use it for that? 0:10:16.679 --> 0:10:18.760 Like you actually use it? What do you use for masking? 0:10:19.240 --> 0:10:21.280 I use voice? I think it's voice AI. I've used 0:10:21.280 --> 0:10:23.040 a couple of them. And by the way, when you 0:10:23.080 --> 0:10:25.240 skim it, you know, when I'm looking for something, it's 0:10:25.280 --> 0:10:28.480 typically at urgency other levels alone, and I can't seem 0:10:28.480 --> 0:10:30.560 to bring them up and over peaky, and then I 0:10:30.640 --> 0:10:33.160 use the service comes back it's too peaky or it's 0:10:33.200 --> 0:10:35.200 got too much base. I don't know, right, So it's 0:10:35.240 --> 0:10:39.160 really about can I use things tools that help get 0:10:39.200 --> 0:10:42.679 to a better artifact than the actual tools that I 0:10:42.760 --> 0:10:45.280 can use, Because democratization of tools means that everything is 0:10:45.320 --> 0:10:47.839 kind of flatlined. So I'm looking for things to become 0:10:48.000 --> 0:10:52.360 really interesting. I don't necessarily I don't necessarily think that 0:10:52.400 --> 0:10:56.640 the quality of the end result is higher resolution enough, 0:10:57.240 --> 0:11:01.360 but it sure beats squeaky markers and poof pads he 0:11:01.480 --> 0:11:04.439 grew up with. So the efficiency of getting ideas out. 0:11:05.280 --> 0:11:08.840 Love that, right, But is it craft if you're using 0:11:08.880 --> 0:11:09.520 generative AI? 0:11:09.800 --> 0:11:11.880 Yes, I think it is. I think it's I think 0:11:12.320 --> 0:11:14.880 an artist can use any tool. This just happens to 0:11:14.880 --> 0:11:15.680 be one of them today. 0:11:16.080 --> 0:11:17.120 Right now. 0:11:17.280 --> 0:11:19.439 You talked about value though, you talked about this thing 0:11:19.440 --> 0:11:21.520 seemed to be bleeding money, et cetera, et cetera, all 0:11:21.520 --> 0:11:24.319 of them. Yeah, God bless, what do you mean, God bless? 0:11:24.440 --> 0:11:26.120 I mean that's that's going to be their problem at 0:11:26.120 --> 0:11:27.560 some point in time. They're going to have to factor 0:11:27.559 --> 0:11:30.000 that out and figure out a model that everyone's trying to, 0:11:30.080 --> 0:11:34.079 you know, generate these tools that can be the panacea, 0:11:34.160 --> 0:11:35.439 and then they're going to make money on it at 0:11:35.440 --> 0:11:40.199 some point. Yeah maybe, yes, See, it's just not all 0:11:40.200 --> 0:11:42.080 the tools he had to make money, you know that. 0:11:42.520 --> 0:11:45.480 Sure, But none of the tools currently make money. Like 0:11:45.600 --> 0:11:48.280 that's that's kind of thing. It just it feels like 0:11:48.320 --> 0:11:52.640 an atypical hype psycle in that having lived through enough 0:11:52.640 --> 0:11:54.839 of these now myself, I'm a little bit younger than you. 0:11:55.800 --> 0:11:58.360 I've never seen one that was just burning cash like 0:11:58.440 --> 0:12:01.319 this and it, but I've never seen one with more 0:12:01.679 --> 0:12:04.040 nor have I ever seen one with more disconnection between 0:12:04.080 --> 0:12:06.959 the utility and the marketing hype. It feels like they 0:12:06.960 --> 0:12:10.440 are promising more than ever. Yeah, and we're in this weird, 0:12:10.920 --> 0:12:13.760 distant area where it's like no one really knows what's 0:12:13.800 --> 0:12:14.400 going to happen. 0:12:15.080 --> 0:12:17.360 Well, you know, I live in the space of ads still, 0:12:17.440 --> 0:12:20.760 so from an ad perspective, there's no shortage of demand. 0:12:21.120 --> 0:12:25.280 But for what though? Ads? Yeah, sure, but demand for ads? 0:12:25.440 --> 0:12:28.360 But what about generative? Like where does general vay I 0:12:28.800 --> 0:12:29.400 fit into this? 0:12:29.480 --> 0:12:31.960 You saying creative but yeah, helping to create the ads. 0:12:31.960 --> 0:12:33.520 I think the ads that are created with Generative are 0:12:33.520 --> 0:12:36.440 pretty rubbish. Yeah, but the truth is there's a demand 0:12:36.440 --> 0:12:39.320 for lots of ads because of just sheer audience growth. 0:12:39.760 --> 0:12:43.040 But the reality is that efficiency of can you make 0:12:43.080 --> 0:12:45.880 ads cheaper? That's a that's a conundrum. 0:12:45.280 --> 0:12:47.400 Because is Generative I doing that? 0:12:47.960 --> 0:12:49.800 I don't think so. I think they're doing some of it. 0:12:49.880 --> 0:12:52.640 Perhaps maybe they're doing an interesting background that would have 0:12:52.679 --> 0:12:54.320 taken taking you up, but it can't. 0:12:54.480 --> 0:12:55.440 It's not the whole thing. 0:12:55.480 --> 0:12:57.080 And by the way, if you and I would look 0:12:57.120 --> 0:12:59.000 at a bunch of ads right now and try and 0:12:59.000 --> 0:13:01.120 pick with their AI or not AI, you would I 0:13:01.120 --> 0:13:03.320 think nine out of ten would get We would know 0:13:03.360 --> 0:13:04.320 that it's written by. 0:13:04.200 --> 0:13:06.439 A There's like an uncanny Valley feel to them. 0:13:06.559 --> 0:13:08.520 It's got a dystopian I guess. It's kind of feels 0:13:08.520 --> 0:13:11.280 a little bit like and we've gone way beyond hallucination. 0:13:11.320 --> 0:13:13.560 We've gone into this kind of you just know, you 0:13:13.559 --> 0:13:15.680 can just see it and feel it. It's cold, it's weird. 0:13:15.840 --> 0:13:18.000 And the funny thing is this, I think what you're 0:13:18.000 --> 0:13:21.079 talking about is not hallucination. It is actually a in 0:13:21.120 --> 0:13:25.400 accurate depiction of something. But it's too It has this 0:13:25.520 --> 0:13:29.679 kind of sheen to it does feels like the movie AI. Yeah, ironically, 0:13:29.840 --> 0:13:31.760 it has said that it came off my head. I'm like, shit, 0:13:31.800 --> 0:13:33.120 I'm stupid anyway. 0:13:32.840 --> 0:13:34.680 And the number of frames a second seem off. 0:13:34.679 --> 0:13:36.640 There's something very there's too much like I. 0:13:37.800 --> 0:13:40.480 Completely agree with you, but it's also where I see 0:13:40.520 --> 0:13:44.520 those sort of things used in more kind of tunny jads, 0:13:44.600 --> 0:13:47.000 like things that are just trying to get a lot 0:13:47.040 --> 0:13:51.440 of these creatives saturating the marketplace, which ultimately if we're 0:13:51.440 --> 0:13:53.920 not careful, we'll make ads more expensive, meaning to get 0:13:53.920 --> 0:13:55.880 to ed the human at the end of it. If 0:13:55.920 --> 0:14:00.720 you if you've now just become completely oblivious or ignoring 0:14:00.800 --> 0:14:03.559 ads that feel like they're aied because and the sack 0:14:03.640 --> 0:14:06.720 gets figured out, right, you'll become blind of these AI ads. 0:14:06.720 --> 0:14:08.280 So to get to us even going to be harder. 0:14:08.360 --> 0:14:11.080 So this is the weird thing right now. Ads and 0:14:11.160 --> 0:14:13.600 AI don't seem to have touched that much. Can I 0:14:13.679 --> 0:14:15.840 say this because there's a company so you heard a 0:14:15.840 --> 0:14:19.320 perplexity too. So Perplexity is allegedly Hayden Field that the 0:14:19.440 --> 0:14:22.360 NBC reported this last year. They're apparently looking for like 0:14:22.520 --> 0:14:26.680 fifty five dollars CPM on what platform on their search 0:14:26.760 --> 0:14:30.320 net their search platform. Okay, but you've not really seen 0:14:30.360 --> 0:14:33.720 ADS and AI touch like even Google's AI search they 0:14:33.920 --> 0:14:34.600 just did ads. 0:14:34.760 --> 0:14:36.880 No, I think perplexity if you were to give it 0:14:36.880 --> 0:14:41.080 a prompt and it recommended something that you were looking for. Yeah, 0:14:41.080 --> 0:14:42.320 I assume you're looking for a piece of gear and 0:14:42.320 --> 0:14:44.800 they recommend a piece of gear versus these other ten 0:14:44.800 --> 0:14:47.240 pieces of gear. Ah, that review is probably in their 0:14:47.240 --> 0:14:50.040 eyes an ad, right, you know what I mean? It's 0:14:50.040 --> 0:14:51.440 not it may not be a display ad as we 0:14:51.640 --> 0:14:52.000 but they're not. 0:14:52.440 --> 0:14:53.680 They're not doing ads yet. 0:14:53.720 --> 0:14:55.240 It's just you know, but I'm saying they could be 0:14:55.240 --> 0:14:59.120 in the description. Sure are they formally not doing ads 0:14:59.160 --> 0:14:59.840 or they haven't. 0:14:59.600 --> 0:15:01.840 They haven't started yet because they want to. 0:15:01.920 --> 0:15:04.480 And when they do those ads, they will be subtle 0:15:04.600 --> 0:15:06.880 like that, they will be within descriptions, they will be 0:15:07.560 --> 0:15:10.360 contextually relevant to that person in terms of the totality 0:15:10.360 --> 0:15:11.960 of a story, not just a display. 0:15:12.040 --> 0:15:13.600 I'm just wanting I think it will be No, no, no, 0:15:13.960 --> 0:15:15.680 I'm not disagreeing with you. I'm just wondering how they 0:15:15.720 --> 0:15:18.080 pull it off. Because the whole thing with hallucinations, you 0:15:18.120 --> 0:15:19.760 can't predict where an AD is going to appear or 0:15:19.800 --> 0:15:20.920 what it's going to appear next to. 0:15:21.600 --> 0:15:23.920 And I think that if it's with with if it's 0:15:23.920 --> 0:15:25.320 with words, the context is wider. 0:15:25.680 --> 0:15:28.040 You think. If it's with voice, they're not going to 0:15:28.320 --> 0:15:29.560 voice as not. 0:15:29.640 --> 0:15:31.680 You know, it's like it's just a wider context. Though, 0:15:31.680 --> 0:15:33.440 So let's let's just stream for a second. If you're 0:15:33.480 --> 0:15:35.800 able to do that, at least you've got a wider 0:15:35.840 --> 0:15:40.840 concept of what the story is versus a display out around. 0:15:41.080 --> 0:15:44.000 But I mean, I mean just like, practically speaking, if 0:15:44.000 --> 0:15:46.720 you vomit out a bunch of text which may rip 0:15:46.760 --> 0:15:49.720 off a journalist outlet, or to how do you like 0:15:49.800 --> 0:15:53.520 I Perhaps the I need to actually ask you the question, 0:15:54.080 --> 0:15:57.000 why do you think that ADS and AI really haven't 0:15:57.000 --> 0:16:00.440 touched it? Because chat GPT doesn't monetize with that, Lexity 0:16:00.520 --> 0:16:03.840 is thinking about it, haven't Google AI has only just 0:16:03.960 --> 0:16:07.440 started considering it. Do you think it's a generative side 0:16:07.520 --> 0:16:08.920 like you've been in the AD slot for that? 0:16:09.120 --> 0:16:12.000 I guess it depends on depends on which place you 0:16:12.040 --> 0:16:14.040 want to sit in that question. Are you talking about 0:16:14.680 --> 0:16:16.800 me as somebody going to the tool to use it? 0:16:16.920 --> 0:16:21.000 Now? I'm wondering why the companies themselves have been hesitant? No, no, no, 0:16:21.160 --> 0:16:21.880 hang on, let me let me. 0:16:21.960 --> 0:16:23.640 Let's get back to the context real quick so I 0:16:23.640 --> 0:16:26.400 can understand the framing. Is it? Is it me going 0:16:26.400 --> 0:16:29.320 to Perplexity's website to be able to use its tool, 0:16:29.360 --> 0:16:31.640 and it's folding in ADS in the results of me? 0:16:32.120 --> 0:16:34.400 I mean, I mean a generative I mean on a 0:16:34.440 --> 0:16:37.160 grander perspective, like taking it away from just like the 0:16:37.240 --> 0:16:39.680 user right now, You've been in ads for a long time, however, 0:16:39.800 --> 0:16:41.520 so I think, why do you think they if it 0:16:41.640 --> 0:16:44.400 was like ads to ad tech has traditionally been very 0:16:44.520 --> 0:16:46.880 quick to adopt stuff. But they will rush. They rush 0:16:46.920 --> 0:16:50.040 to fucking micat, They rushed to everything fast. Remember that. 0:16:51.480 --> 0:16:55.040 But it's AI. They've generative AI, generative search Chat GPT. 0:16:55.240 --> 0:16:57.440 Because I just don't think it's I would I would 0:16:57.520 --> 0:17:02.160 say that it's just a little naive. And I have 0:17:02.240 --> 0:17:06.240 seen you mean, I have seen ads that are directed 0:17:06.320 --> 0:17:13.320 by so, scripted by AI, shot by humans. Sure, and 0:17:13.440 --> 0:17:17.600 they look there's just something missing. And because that litmus 0:17:17.640 --> 0:17:19.480 test of is it warm and after does it feel 0:17:19.680 --> 0:17:23.359 like something that is a motive it misses? Sure, it 0:17:23.440 --> 0:17:25.239 is going to get a lot better. Man, This is 0:17:25.240 --> 0:17:26.360 not even a debate, but. 0:17:26.600 --> 0:17:29.040 I mean it absolutely is a debate. I mean I 0:17:29.080 --> 0:17:31.399 will absolutely debate the shit out of this. The training 0:17:31.480 --> 0:17:34.320 data required to make saua better does not exist, even 0:17:34.359 --> 0:17:37.080 if you took every video ever taken there. Adobe is 0:17:37.080 --> 0:17:40.320 paying people to take video. But my question was actually 0:17:40.359 --> 0:17:43.639 really way more specific, which is, you have been in 0:17:43.680 --> 0:17:46.680 ad tech for a long time. Everything else has been 0:17:46.680 --> 0:17:51.800 in like generative AI platforms are not integrating ads as 0:17:51.840 --> 0:17:55.200 a monetization mechanis well, and why do you think that is. 0:17:55.000 --> 0:17:57.920 Because the presentation layer of that, the actual ability to 0:17:58.040 --> 0:18:01.159 generate something that feels like it's not coal specifically, and 0:18:01.240 --> 0:18:03.439 I've said it a few times now, is not up 0:18:03.440 --> 0:18:06.320 to snuff. Everything else in the background that is actually 0:18:06.440 --> 0:18:10.600 using lots of big data to be able to represent 0:18:10.640 --> 0:18:13.320 the right type of context to you. Today we will 0:18:13.320 --> 0:18:16.000 call that AI is in play. So I think there's 0:18:16.000 --> 0:18:18.359 a marriage between what's going on in the back Haman 0:18:18.400 --> 0:18:21.320 along and the presentation layer of that, to be honest, 0:18:21.440 --> 0:18:24.320 is terrible. I agre in comparison, I agree, but maybe 0:18:24.320 --> 0:18:27.120 they need to be more specific. It feels like ad 0:18:27.200 --> 0:18:29.840 tech is they will do the generative side. There's tons 0:18:29.880 --> 0:18:31.760 of ad tech platforms that will generate like and you 0:18:31.760 --> 0:18:34.600