Become a Creator today!Start creating today - Share your story with the world!
Start for free
00:00:00
00:00:01
Mind the Model: Episode 004: Paul Blackburn
35 Plays21 days ago
This week we're featuring our first live episode recorded live at SXSW Sydney in collaboration with Paul Blackburn, Chief Growth Officer at Vudoo, and one of the co-hosts of Commerce Media Matters.
Be sure to check out…
- Their podcast: https://vudoo.com/commerce-media-matters-podcast/
- Paul’s LinkedIn: https://www.linkedin.com/in/paulblackburn/
Recommended
Transcript
Live Collaboration Introduction
00:00:09
Speaker
Hi, and welcome to a very special live collab of Mind the Model and Commerce Media Matters from the team at Voodoo. People, I know most of you are already sitting down, which is good because I think this is podcast gold. This is you know, you're go your children will be talking, you'll be telling your grandkids about the day that you saw this collab happen live here right now at Side by Southwest. It's big. It's big.
00:00:31
Speaker
This is, yeah, so I'm glad it's being recorded. So first of all, I'm Nathan Guerra, and with me as always is my podcast co-host, Emily Crowlin. Howdy. And I'm Paul Blackburn. um I am the chief growth officer at Voodoo, but also the co-host of Commerce Media Matters. I've been allowed out on my own, um so without my co-host, he sends his apologies, Nick, but he's not very clever anyway. ah Can we edit that for later, please?
00:00:57
Speaker
but No. That's staying. There goes my job. Yeah.
AI Models and Opportunities
00:01:01
Speaker
Super exciting, though, because both of our podcasts really are and talking about the the advent of ai innovation, creativity, commerce media. There's heaps to cover off today. and But just wanted to start it off the way we usually do on ah Mind the Model. So, Paul, tell us what's your favorite AI model and why? And what was the last thing you used it for?
00:01:23
Speaker
it's I've got a lot of favorites, but the one I use the most is ChatGPT. Just started there, invested so much time and effort into it, it's now hard to change the model. um and ah yeah and the And the last thing I use it for, I think we were talking about this in the break, was I'm selling my dad's furniture.
00:01:43
Speaker
So I took photos, loaded it onto ChatGPT to ask what furniture it was first, what it was made out of, and can you make a Facebook marketplace listing that sounds cool out of it for me? And it did, and it sold really well. So that's the last time I used it.
00:01:58
Speaker
Very good. What about you guys? And what have you done recently? I mean, it's nerdy, I suppose. But after we launched the podcast on YouTube, I put in into Notebook LM, and then I asked it to come back with the perfect caption for both LinkedIn and Instagram.
00:02:13
Speaker
yeah So on the bus ride down here, i was trying to put on more Instagram content. so it was, you know, quick captions and referencing exactly what it was used for. Yeah. I was writing some JavaScript.
00:02:25
Speaker
So, yeah. I don't know anything about JavaScript, but I was writing it so with ah the help of Gemini. Well, it sounds like you guys are using it for good, not evil. My son, however, is using it to cheat. So he's writing book reports, but he's learned prompt engineering pretty well. So he firstly prompt engineers saying, write the book report so that the ah the AI detectors at the school can't pick this up. okay So that's his most recent usage. Yes. Perfect. Well, you know, we all use it for the things that are important to us. Yeah, exactly. um All right. our um Our other question we'd like to ask people is, are you an AI optimist, a realist, or a pessimist?
00:03:00
Speaker
I'm definitely an optimist. Okay. Why? i want to oh oh Because you've got to be. You've got to be. um I think there's, you know, there's real opportunity out there. i think that, um you know, we've got great people. If anyone saw Mo Godot speak earlier in the week, um you know, there is an opportunity here for humanity. There might be some, you know, damage on the way and we we'll do some learning on the way. sure But I'm an optimist around the fact that it's going to create new jobs, um not replace them, particularly if people lean in. I'm an optimist because I think that it it will provide in you know, maybe ah places where they don't have the same advantages for education or democratize information a bit more even further than the Internet does.
00:03:41
Speaker
um But, yeah, we're going to face some real problems at the same time. But I'd like to be optimistic that we can get through those problems. I don't think we're going Terminator. so And AI will solve those problems
Human Creativity in AI's Future
00:03:51
Speaker
as well? or I think people using AI will solve those problems. I still think that it's AI is a tool. I think humans, you know, we were talking about this actually. It's like, what do you tell your kids around to focus on for their studies in and and a world of AI?
00:04:07
Speaker
Where, like we were talking, you used to learn code at and and so forth. And now people can just vibe code. ah Totally. That's what I was doing earlier. like That's right. I think we can train our kids or tell our kids to have inquisitive minds, to ask questions, ah to learn, to to to sort of delve deeper with the answers they get from these tools. And they and and humans humans' minds are so creative. I'm not sure...
00:04:31
Speaker
I'm not sure there's a, ah hopefully there's not a time where that can be replaced. You know, it's human instincts, human emotions. um You know, will the machines ever get to that point where they they probably can model them?
00:04:43
Speaker
But I'm quite optimistic that there's a role for humans. There's a role for humans in the future. think that's what we have to kind of turn to because if we're not hopeful about the opportunity for AI, like we we don't want to think the world is going to end tomorrow based on, you know, AGI coming and, you know, terminating all of us. Like, I think that's why i I find myself to be in the optimist camp, but I'm also a bit of a realist because to be honest, like it's, yes, it's advancing, but we're also at South by Southwest in Sydney, like very tuned into the tech and innovation. yeah um I don't think the adoption is widespread enough. It's definitely increasing, but it's not widespread enough to completely advance every single person's life. I think there's a bit of a growth to, to happen.
00:05:27
Speaker
Um, So that's why I'm more of optimist realist at this stage. Okay. Yeah. I mean, ah I mentioned my son he's ah and and and my daughter. They're a great test for how things are going in the future. you know they but They've been able to use you natural language, as we were talking about before, yeah um to do things. It's great for them.
00:05:45
Speaker
um But on the flip side, and to to back up your point, I use my mom and my dad, and they're not using AI much anymore. So like you said, it's not... any Anymore before they were using it before? Sorry? You said anymore. Oh, sorry. I didn't mean to say anymore. Okay. i I was like, well they were using and then they rejected it? No. way Yeah. Yeah. yeah yeah he he'd hed He'd think that's somebody's initials.
00:06:05
Speaker
You know, like, who's who's Alexander Intelligent Man? Yeah. and Yeah. yeah i'm All right.
AI's Impact on Commerce
00:06:11
Speaker
Cool. Let's jump into some of the ah commerce stuff. I'm really kind of curious because ah we saw Shopify and now... This gets me excited, the commerce space. Yeah. yeah yeah We're getting in. yeah ah Walmart announced yesterday that they are now integrating into ChatGPT.
00:06:27
Speaker
The beginnings of AI, agentic shopping. Do you think this is improving the consumer experience right now? Look, um it's a great announcement. We're really excited about it. are From a voodoo perspective, that's kind of ah the you know the dream that we see is that you know you should be able to trans ah transact from the point of inspiration.
00:06:47
Speaker
um And i think I suppose now you're in a GPT environment, right now you do all that conversational piece to find the, we've we've got a great case study that we ran where it was like finding a dress for a cocktail party yeah that we we we ran a sim on.
00:07:02
Speaker
And you do all that work to find the right dress and it tells you, but then you can't transact there. So you go you sort of it it hit a dead point. You've got to go open another browser or whatever else. Announcements like, um you know, Shopify, Etsy and and Walmart now being able to transact with inside. It kind of meets the the the UX that the consumer is expecting. Makes them severe to catch up with you as well. Yeah. It's also like it's a tool and I think consumers, you we go to think about a consumer. that's They get to this end point and now we're allowing them to purchase and and and and and so forth. So I'm really excited about what this means for for for for the future.
00:07:36
Speaker
Also for monetization. I mean, you know, you've got all these VCs pouring a heap of money into the dream of what AI is going to do, but do they have a model that's going to show a positive revenue? soon Sooner or later, the VCs are going to go, where's the money? Yeah.
00:07:53
Speaker
Where's the money and how where's the road to revenue? And it can't be just on subscription. I mean, these models, the more advanced they get, they just cost so much money to power the rest of it. So there has to be other ways to create revenue.
00:08:05
Speaker
um And I think we heard Sam Altman say, you know, it's when I don't think it's going to be straight ads. And so the affiliate model, the shopping model, the service to the user mo type models are the ones that I think will can be really interesting. Certainly why Voodoo exists and why commerce media, where commerce media matter thinks they they it should
Retail Media Power Shift
00:08:24
Speaker
go.
00:08:24
Speaker
Right. And so is that's about reducing friction at the end of the day. Absolutely. Okay. Yeah. I mean, ah like you ask it for a recommendation and it gives you it and you can now just click to buy it. I mean, it's consumers are already educated this way. They're seeing this in social media at the moment, right? Yeah. yeah We had the influencer marketing guys talk about that.
00:08:42
Speaker
You've got the influencer saying, click here to shop now. it's It's the expectation of a user now that where they get their recommendation from or where they, where no matter where they are in the in the purchase journey, if that last point's where they're inspired, they want to be able to shop.
00:08:56
Speaker
Okay. Yeah. Yeah. Fair. Yeah. Well, I'm just thinking about a feature where, and bear with me on this one. So let's say, for example, retailers in Australia start to integrate within ChatGPT and you can click to buy within that user interface.
00:09:11
Speaker
I'm a bit of a data geek myself. The transaction data, and the first party data would be held by both OpenAI because it was done within their platform and the retailer because they would recognize the revenue at the end of the day.
00:09:22
Speaker
So now you have two distinct and companies holding the thing that is so valuable, which is first party data. What does that do to the future of retail media networks? Does it dilute the value of retailers who do decide to integrate within and platforms like ChatGPT?
00:09:39
Speaker
it's ah It's an interesting question. It certainly rebalances the power. Absolutely. um If you think about even with those announcements we talked about before, so you've got Etsy, Shopify and Walmart, and now OpenAI has a view across all of those things. That's that's a pretty powerful ah proposition.
00:09:56
Speaker
Yeah. So I think it sort of rebalances the game back in the power of these big LLM models and so forth because they have that data set. um But I don't think it it it limits the retail media guys with their powerful data sets to do stuff as well.
00:10:10
Speaker
So you think they're getting all the first party data? So look, ah without trying to talk about the product that that we we sell at Voodoo, but to use it as an example, the the retailer is still seeing the transaction data too.
00:10:23
Speaker
yeah So what they're not seeing at the moment, so they've got an advantage here too. So Walmart will see the transactions that happen inside Voodoo. The GPT. So they right now, that's a blind spot for them.
00:10:35
Speaker
They're not seeing anything. They're just seeing referrals coming from there, and that's it. They're not seeing what's happening in there. So they're going to get some new data points as well. They're going to get the prompt engineering that then triggered the ah the transaction and other bits and pieces. So I think there's data in it for for the retail media networks as well.
00:10:53
Speaker
And then, you know, the about... that the the thing about the The commerce media ecosystem is how do you unlock that data to find ways to talk to a consumer at the right time?
00:11:04
Speaker
And it's not always going to be in the LLM because you're going to need to have a brand conversation with them to, you know, how do you how and how does a brand show up in an LLM now? Yeah. Yeah. I know a lot of marketers are talking about that, about like the change of consumer behavior. And, you know, not everyone's just only going to be in ChatGPT talking about ah the product that they're going to pass by out of home placements. They're going to be watching TV or, you know, streaming or, you know, having conversations with friends following their favorite influencers. I think there's such ah demand for how brands will show up.
00:11:36
Speaker
outside of that performance space and more in the branding awareness. And I think personally, that's where retailers are really going to excel because at the end of the day, 17% happens on the online e-commerce experience when, you know, a majority still happens in a retailer.
00:11:51
Speaker
And I think in in a world of AI for a brand, you're going to need to show up in the real world a lot more. For sure. To get that people preferring the brand. And that's where retail media ah with their their bricks and mortar shop front have an opportunity on that consumer journey to then have brands show up. That's why I think retail media is really interesting because you're already going to be buying something from that retailer.
00:12:13
Speaker
And on that journey, you can have another brand show up to talk to someone. So it's a really interesting space for brands. But I think retail media has got to learn how to enable brands to show up in their media. I think we're at the start of that right now. It's very transactional. It's very like a listing.
00:12:28
Speaker
doesn't allow for a brand to communicate their values like a publisher site does. And I think yeah e retail media could learn a lot from the publishing world around how um yeah to allow the endemic and non-endemic rep brands to show up and actually screen their brand values. And that's with great content.
00:12:45
Speaker
And is anybody doing it well? um Not sure. our Question answered. Okay. Yeah. and yeah I mean, um it's early days. I mean, but most of the big retail media networks have just got to the point where they've got the infrastructure to actually behave like a publisher. yeah There's a lot of great people, you know, Qantas does well. Chemist Warehouse does well.
00:13:10
Speaker
The House of Wellness is a brand that the Chemist Warehouse publishes where they actually publish great consumer content. It also sits within Body and Soul, which is a News Corp website. So their content sits as consumer content.
00:13:23
Speaker
they And they do a very good job. They rely on a business called Medium Rare, who are ex-journos, ex-magazine, ex-digital content producers to do it because retailers don't do lifestyle content very well. They think like a retailer.
00:13:37
Speaker
um So, yeah, I'd say Chemist Warehouse would be one of the ones that are doing it quite well. Okay. We talked a little bit about brand, but when I think about agentic shopping right now, it feels very much like from an advertising perspective, you're talking about performance media.
00:13:51
Speaker
I mean, is is this taking money from Meta and Google or is it going to be raising the entire ecosystem?
AI and Advertising Revenue Shift
00:13:57
Speaker
Or what's what's your thought in terms of where the money is going to flow? Initially, they're going down the performance route because it's about are getting to that transaction and that's how they're going to attribute where the money flows.
00:14:08
Speaker
yeah um I think what we're learning is that there's a lot of value outside of just the transaction. So the interaction, to the look that when you're ah not ready to purchase, you're just starting out your journey and you're able to to click on a piece of content and and and go through the catalog, look at the colors, look at the availability.
00:14:27
Speaker
They're all great data points to then use to to retarget and and ah they're they've slightly further up the funnel. um So yeah, initially it's always about performance because no market has ever been fired for being able to go I spent that dollar and I got $50 back. Yeah.
00:14:42
Speaker
But brand is harder to justify. So I think there's some great upper funnel opportunities, particularly with the data that you're going to get. But yeah, I think originally it would be it would to be performance. So Meta will embrace it because they are they're great at doing performance.
00:14:56
Speaker
Yeah, but I mean, people are going to be putting money into ChatGPT. 100%. They will take it out of meta. They will take it out of Google. I think there is going to be a bit of a rebalancing of power. And OpenAI i naturally sees advertising as a revenue raiser because, you know, we love making money in advertising.
00:15:14
Speaker
But I don't know. I think... Especially with you know the development Sora and the pivoting towards trying to make a social network. OpenAI, i from my perspective, is trying to be that one app for everything.
00:15:26
Speaker
And I think to your point, yes, they are going to go down the route of performance first because it's a natural fit for them. But it's only a matter of time before they open up shop. I mean, they've already opened up shop in Sydney. It's only matter of time before they actually start to put ads within the interface itself. And it will come, I think, first all.
00:15:44
Speaker
from Google because it's a natural, you know, asking. And then it will also come from Meta because it's just the black box that is known for performance. That's my take anyways. I mean, we it's easy for Google. I mean, they've got a great search business. Yep.
00:16:01
Speaker
ready to go to put in and supplement your conversational pieces within a GPT and then suggest some other links in there. yeah Same with Microsoft, with Copilot and the rest of it. they've got ah They've got an ads business that would naturally fit within there.
00:16:14
Speaker
My hope though is that OpenAI does it better than that. And then we reinvent what ads are that are a bit more friendly. They're not plastered all over the internet. And it's a bit of a cleaner environment. More tailored to the consumer as well.
00:16:27
Speaker
Say again? More tailored to the consumer because they hub know so much about you in the context of who you are. Like, you know, if you've ever asked ChatGPT or your preferred LLM, what do you know about me? And then it comes back with what it thinks about you. You start to get scared.
00:16:40
Speaker
Oh, you're pretty on point with you have a really good understanding of who I am based on how much I ask of you. So think about what advertising would look like in that kind of ecosystem where it could be especially tailored for you.
00:16:52
Speaker
But I mean mean, you know, having worked at Google for 12 years, Google knows a lot about you. Trust me. And that's because nope I always said nobody ever goes to Google on lies. Right. You don't go to be like, I'm going to fight. I'm going to play with Google today. Ha ha ha.
00:17:06
Speaker
and No, you go in and you ask a question because that's something you care about me some embarrassing questions. that Absolutely. Things you wouldn't ask your friends and neighbors probably. That's right. I mean, look, I hope we're at an inflection point where OpenAI can clean up this the advertising ecosystem. And it's ah it's a it's a once in a and and a lifetime opportunity to kind of to to clean up some of the things that we we did wrong with the first advertising phase of the internet, um um where I think it's you know that that clutter It's just everywhere. I mean, it's... My picture on that, though, is like, and and you've seen it with Google, is you start off very simple, and then there's an opportunity to make more money by adding another ad spot, by adding another ad spot, and shareholders want more value. And so that just becomes... Start soft clean and gets more and more cluttered over time.
00:17:51
Speaker
And that's exactly what you'll see with ChatGPT. Yeah. It'll be a race to the bomb. Like they want to make as much money as possible. Like you mentioned earlier, it costs bananas amount of money to run these LLMs. Yeah. It's only inevitable that they're going to try and squeeze it for everything they can, whether that be through subscriptions or, you know, integrations, partnerships, referral, all of the above. Yeah. Yeah.
Copyright Concerns and Adaptation
00:18:13
Speaker
You know, one thing we touched on previously, and I know I'm a bit passionate about and you as well, based on your background, is talking about premium content. And, absolutely you know, the LLMs, they were trained off massive data sets online, not all of which that they asked permission and paid for. So a lot of it's. That's right. Quite passionate about that point, actually. Copyrighted material. um One of my favorite podcasts, The Daily, or sorry, not The Daily, The um Hard Fork from The New York Times. they always One of the hosts always has to start with a disclosure saying, I work for The New York Times, which is suing OpenAI and Microsoft for copyright infringement.
00:18:46
Speaker
Just as a reminder, like sitting in the States, we're talking about AI, but we're taking them to court as well. and What's your take on this and what happens to the future of that written news and premium content when AI can just summarize or repeat it and spit it out within the box that you're interfacing with? That's a huge question. I reckon if I knew it, I would be worth a lot of money to the publishing world. But ah look, for me, it's an interesting move to see that OpenAI has already seen the virtues of the crap content that's out there. um They're leaning towards more qualified, journalistic, triple-checked content. They've they've deleted Reddit from some of it you know um they add sources its... It's sources because that's where a lot of the hallucinations are coming from, from UGC content.
00:19:32
Speaker
I think premium publishers, i don't know what it looks like, but they have an opportunity to play the trust game. you know they've they've they've They've got trusted journalists that have worked in you know You take Vogue, for example. I mean, it's very hard. that you'veve You've trained your whole life around fashion, and fashion is a really difficult subject to have ah an opinion on. you You finally get it to Vogue. You've you've spent your whole life on that.
00:19:56
Speaker
yeah When you're going to advice, you want to go to someone like that. okay ah You don't want to go to the open internet and and and things. So I feel like the the publishers have an opportunity here to play the trusted advisor game, yeah um but they need to disrupt themselves.
00:20:10
Speaker
They need to be having ah ecosystems that are like conversational commerce. you You know, you need to go to Vogue and be able to talk to Vogue like you've talked to ChatGPT. Yeah. um So they the the the the publishers that sort of embrace that, I think they have an opportunity to play the trusted advisor game.
00:20:28
Speaker
And, you know, look you look at back at they the amount content that they've got that's been sourced. Yeah. like News Corp or not, or like fairfax of the old Fairfax 9 Publishing, sorry, now or not, they are held to a quite a high regard. They have to have their facts checked. they There is laws in this country that doesn't apply to the open internet. no So the information can be trusted above the open internet. So yeah they've got an opportunity to help train models that you can trust.
00:20:57
Speaker
But it's assuming at the very best worlds, not all of open AI is going to be trusted content from brand partners. No. and and And how do these our premium publishers make money in the world of open AI?
00:21:12
Speaker
I mean, they've obviously, some of them have got the opportunity to go to the negotiation table with Sam Altman and they're yeah they' they're getting paid a little bit. But, you know, i look at Google, Google, who work with publishing models where the publishers have an affiliate model, where they were doing reviews of products and Google would advise you need to get longer and longer reviews. You need to do more editorial coverage to get that view to appear up in the search result to then get you, get them in top of funnel to then make their affiliate revenue.
00:21:40
Speaker
And then they launched AI overviews yeah and took 40% of traffic off all publishers internationally and just absolutely ruined that business model for them. so So how do we get the money back to the original publisher of the content that did all that hard work is something that I think, you know, that's one of the things that we need to to grapple with. And how do you how do you how do you trust the content that's in these models is is something we need to grapple with. Yeah.
00:22:04
Speaker
And I don't think there's a good answer. There's a problem, right? Yeah, yeah, yeah. Well, i can assume i I can assume what your answers answer will be, but you know we have different tech leaders who are coming to the table to say, let's let AI train on all work.
00:22:18
Speaker
The line between innovation and exploitation, we have an old mate, Scotty Farquhar, who's you know had a bit of an atrocious interview on the ABC talking about how copyright materials shouldn't stand in the way of innovation for ai And then you know Michael Miller's rebuttal recently going... My old boss. Yes, going to talk about how, no, it actually needs to be respected and the copyright is there for, you know, um bringing back to the, don't know, just supporting the premium content and the journalistic standards.
00:22:46
Speaker
It's the value. Yeah, the value. The value of the content. What's your take? and is there a line that should be drawn or do you think that we could kind of waver more towards innovation if it means the models get trained and better and guess more trusted content comes from them because they're, you know, training on that premium content?
00:23:04
Speaker
I like to say there's a room for both. I think, um you know, in some of the publishers' negotiations with OpenAI, they're allowing snippets, which is maybe enough to keep doing the innovation.
00:23:16
Speaker
So you do not want to stymie innovation, but you've got to think of the cost that it costs these publishers to create this the the the the original content. I mean, if you look at some of the stuff that News Corp has done recently with the the Teacher's Pet podcast,
00:23:33
Speaker
But they actually changed laws for domestic violence because there was an investigative piece of journalism that yeah found that a man got away with murdering his wife.
00:23:44
Speaker
That costs a lot of money yeah um to do. and And so you've got to be able to find a way to fund this. It's got to be a payback. You can't just have ChatGPT come and take all that information and surface it for someone. There's got to be a payback.
00:23:56
Speaker
So there's a happy medium. um And I think Michael Miller, knowing him well, would support the innovation piece. But at the moment, it's probably skewed in the tech giant's favor. It's what his probably his point is.
00:24:08
Speaker
ah How do these premium publishers monetize through these new models? that There's not a clear path for that. Yeah. It's definitely a brave new world out there. It's yeah. Yeah. No one's figured it out to your point. Like everyone's just trying to figure out whether that means taking a stand but against any kind of partnership or getting them early to have some kind of money flow back through to them.
00:24:27
Speaker
I have a big, bill I have a, and there's a mix, right? There's the people that have jumped to do a deal and there's the people that are standing their ground. i mean, I'd like to see a cost per citation type model. Oh wow. Okay.
00:24:38
Speaker
You know, I think there's, that there's some grounds for that. um Then who pays for that? the ads or the the the money that comes from these massive audiences that are now going to chat GPT instead of Google or instead of a pre-owned publisher. So almost like a so YouTube kind of revenue share model then?
00:24:53
Speaker
ah Something like that. I mean, hey i don't know I don't have all the answers, but I think it's ah it's a great way to start. I mean, it means that the... the the um the the consumer is aware that the that you know if you're citing citing an original piece of content, that it's that they can trust it or they can or or they might not trust it. yeah So the citation becomes an important part of the game so that you know where the content originates from.
00:25:18
Speaker
i think that's important for a consumer as well as the original content producer. And if you know some of them some of the LLMs, like a perplexity of the rest of it, can then work out how to pay and They won't be being sued by the original copywriters. So it could be a win-win for the ecosystem if they if they can work out some sort of model that works that way.
00:25:37
Speaker
Okay. what What about, you I mean, what do you guys, you know, how do you how do you see these models giving revenue back to the people that originally created the content?
Monetizing Content in AI Era
00:25:47
Speaker
ah It's an interesting one. So I don't see them giving back.
00:25:51
Speaker
I mean, I don't see them, ah frankly, like knowing how they work, I don't see them having a a really clear, easy way of doing it. It's not like YouTube where you can go, this piece of content I'm showing, and so we can split the ad revenue on it.
00:26:02
Speaker
Yeah. I don't see them volunteering and I don't see it happening on an easy will level. So do you think we'll see a lot of lobbying of governments to get behind copyright protection? i i don't really think government's going to step up in the US at the moment. I would hope that Australia does.
00:26:19
Speaker
I just, I ah would love to see government lobby against the... um the copyright infringement and how the models are being sourced currently.
00:26:30
Speaker
i just don't think that's going to happen. i'm I'm aspirational about it actually coming through, but I don't think it'll happen. Yeah. I mean, the other thing I see is, don't know if you guys agree, but the fight back from the publishers is they're employing people that understand these models so they will cut the information off. But like you said, it's going to take a lot of media warring media companies to sort of operate together without colluding.
00:26:54
Speaker
There's rules around that too. Of course. To stop these models from using their content. So it's it's definitely an interesting space. Yeah. I mean, I think licensing is probably the way forward. yeah And I think that's where the copyright rulings in the States may play out. Like you may very well say, okay, New York Times owns that content and they have the right to say, no, you can't have it.
00:27:13
Speaker
And I then think that you're going to see licensing deals for that. But ah what I think that's going to mean though, is you're going to continue to see smaller and smaller, sorry, larger and larger media organizations and a smaller number of them because small organizations won't have the ability to bargain.
00:27:29
Speaker
Like the New York Times will. Yeah. Or like the News Corpse of the exactly the or the Condé Nast or the big guys of the world. Yeah. know and and then and it And it comes back to also monetization again. Totally. to To bring it back full circle.
00:27:40
Speaker
Because if there is new ways to monetize content for the LLMs, then there's more money for them to then look at models that effectively pay back publishers or do deals directly with the second tier or third tier publisher out world. Yeah. Yeah. But it's ah it's an it's an interesting game.
00:27:58
Speaker
Do you think the pro models will then eventually be made cheap enough that they'll be advertising funded? Like you'll get that on your homepage unless you want to pay for the subscription? Is that where we're headed?
AI Revenue Streams and Competition
00:28:10
Speaker
I think eventually there'll be an ad-funded model. um I mean, we've seen it with everything, haven't we? I mean, I remember when I first got Foxtel and the whole sale was no ads. So I paid a subscription fee.
00:28:21
Speaker
Then all of a sudden there was ads. You know, I mean, Disney Plus and now there's ads. You know, like, hang on. with' we So, look, I think we've seen through history that subscription-only models aren't able to sustain.
00:28:34
Speaker
you need more revenue streams. Eventually, advertising comes for all of them. Well, advertising, affiliate revenue models, other revenue streams need to be there to to maintain it. I mean, you've got other countries like China that don't have the same sort of challenges around funding because the state's funding their developments and it's scary how fast they're moving.
00:28:54
Speaker
So the American-based tech companies need to find the money and the revenue and the rest of it way forward so that they can continue in the arms race of who's going to have the better models.
00:29:04
Speaker
Yeah, although they have the additional challenge of a government which is much more concerned about the size of those organizations that the Chinese government don't have or know ah companies don't have to reckon with. that's That's right.
00:29:16
Speaker
That's right. A friend of mine has just, he's in the AIS space. He's just come back from China though. And he was extremely surprised how far ahead they are. um And we're not seeing it here. No. um But they are, this is, they're really going ahead. So the AI guys need to find new revenue streams so that they can advance and to keep up and not fall behind um yeah from a Western world perspective.
00:29:41
Speaker
Yeah, I mean, you can, with the recent news of the um the deals with NVIDIA, the chip maker, um i think the other one was AGM, might be messing up those letters, but um OpenAI doing direct deals with chip makers trying to get billions and billions of dollars to fund the innovation, fund the um the compute power that's required, the amount of energy and water cooling. Like, it's it's absolutely mad how much money is being pumped into this. But I know that They're all trying to race to the top. They think it'll be winner take all once they get to a an AGI state where computers are smarter than us.
00:30:16
Speaker
And I think it's the way they're approaching it is, you know, at all costs, innovation at all costs.
Ethical Guardrails for AI
00:30:25
Speaker
But, you know, on a previous panel, Nate and I saw the other day, one of the speakers said, well, at what costs?
00:30:32
Speaker
Who's going to fail in, you know, if there's not the guardrails for safety, if yeah people are you know, interacting with LLMs and um being advised to take their own life. Like it's,
00:30:43
Speaker
There should be some level of regulation, but not to stymie innovation. So I know it's a hard one. Clearly no one's solved it because we're all still talking about it. This is pretty much the theme of South by Southwest this week.
00:30:55
Speaker
don't understand. And look, the the Internet is a reflection of humans. And it's a messy, yeah horrible space. And these models have been trained on on that. And I you know i want to go back to Mo Godot, who spoke earlier and at this conference.
00:31:10
Speaker
Where I'm an optimist is that we know this. And we can if we if we act now to create great guardrails, and we all have a voice in this, every every consumer has a voice in this, every organization has a voice in this.
00:31:21
Speaker
If we can start to think about how we talk to each other, what we put on the internet, create the guardrails, think about the ethics behind all these things, you then we we we we might actually stand a chance for some some some a rosy future. Positive outlook.
00:31:35
Speaker
That optimistic piece is coming out. you know To and completely swap lanes from the very hopeful optimistic viewpoint, let's talk about AI
Authenticity Issues with AI Videos
00:31:44
Speaker
slop. Oh, yes. i yes yeah So currently only available in the US, but OpenAI just recently released Sora 2, which is AI generated video. Facebook has Vibes. I don't know who's in charge of creating that name.
00:31:58
Speaker
It is start is quite scary to see the videos that are coming out of this where, you know, they're fully AI generated. I saw one and that, you know, was reposted on Instagram last night and it was a baby being born and then the baby ran off.
00:32:12
Speaker
down the halls of the hospital, and it looked real. And it's one of those, like, this is bizarre. Currently, it's being used for, like, you know, fun videos of friends or, you know, let me have a conversation with Sam Altman on my podcast.
00:32:24
Speaker
and But what's your take on the AI sloth? Well, I mean, again, coming from a premium publisher, it's pretty scary because to create a video of the standard that you can do with a couple of prompts costs a lot of money in a publisher, a lot of people power, lot of technology, lot of subscriptions to Adobe, ah you know, that sort of stuff. So it's's it's scary proposition from there.
00:32:48
Speaker
um I guess copyright is this there's another, this this is where I think it's really interesting. I think in your space is like, how do we get copyright in this space that you know that it's, because I'm scared of how good the fakes are. I mean, I'm pretty savvy, um but I was just talking about my 78 year old parents.
00:33:06
Speaker
They're not, they're seeing videos that they think are real. um And this is not my kids seeing them. This is the parents seeing them because they are so accurate and and real. And it's scary how easy it is to put out communications, to put out videos that seem real, that are fake news. It's an epidemic.
00:33:23
Speaker
Well, and there's a whole like stream of content on TikTok and Instagram of people's parents saying, have you seen this video? And them going, mom, there's no 10 foot tall polar bear in Florida fighting gators.
00:33:36
Speaker
It's not happening. And people people are constantly confused by it. Yeah. Yeah. A hundred percent. So ah don't know what the answer is, but again, I think we, we, we need to lobby around guardrails. Um, we need to, to, to have the conversations around how we make sure that people are educated around, around this, because, but you know, I grew up in a world where you went, you turned on the TV and you watched the news and you trusted the news.
00:33:59
Speaker
Yeah. And that was the only so people that could create content like that. Cause it was live content or, or Hollywood at the best, but you knew it was Hollywood. Yeah. Now it's, it's all blurred. Well, do you think it's a passing trend? Like, oh, it's open AI trying something new. It's, you know, meta trying something new. Or do you think it's it's around to stay where consumers can produce something that looks so real and is so visually engaging that it kind of blurs the reality? I mean, how can it go back?
00:34:28
Speaker
A hundred percent it's here to stay. I mean, there are some optimists in this space. So David Droger, it can spoke around the fact that, you know, he's a creative and he's excited because he says these tools will allow them to be even more creative. yeah There's still original thinking that can be synthesized using these tools yeah ah to make even more entertaining content and the rest of it. yeah That's the optimist view. the yeah and And they might both be right.
00:34:54
Speaker
I mean, we might be constantly confusing everyone and nobody believing anything and still being able to make things that are incredibly creative. And some of the stuff is very funny and entertaining. and Don't get me wrong. I mean, yeah I watched one around a yeah ah ah a dinosaur picking up a baby. It was really quite interesting. it was really quite funny. Yeah.
00:35:14
Speaker
Yeah. but but it just It goes back to that brain rot. I don't know. I think it's a fad. I think it's funny to watch. But then at the end of the day, I think back of my... like and you know I'm not on the the AI slot apps.
00:35:26
Speaker
But when I do go to Instagram, I want to watch videos that are trusted and real and not fake stuff. Because it's funny. But then it's like, okay... You know, we got the lulls.
00:35:36
Speaker
Then what? it's I just think it's then and it's the cheap French fries of the Internet. It's not going to stick around. That's my take anyways.
AI Content as Entertainment
00:35:45
Speaker
Okay. Well, I mean, i I'm with you. I like to know what I'm watching is real, but there's a market for it.
00:35:49
Speaker
I mean, yeah ah you've only got to look at people watching the slop that's already been on YouTube for a long, long time. And people just sit there watching it, entertain, thinking it's real.
00:36:00
Speaker
yeah Or they don't think it's real, but they're entertained by it. And so it's just proliferated that. It's entertainment. You're exactly right. yeah And, you know, you bring up the good point earlier about the copyright. That's something that we discussed previously on, you know, train the models. But then also it's the the production.
00:36:17
Speaker
You know, you have all the big Hollywood execs who are starting to say, hey, wait a second. You're, you know, that looks like it's created in the likeness of Garfield. No, no, no. We have strong copyrights in place for that very reason.
00:36:28
Speaker
But yeah, it is entertainment, but I think it's garbage. It's in time. And I think then there's also going to be this trust issue. Like you said, like where, how do you trust? So then are you going to search out, you know, a Disney or a CNN to then view their AI content? Because at least they're a brand that stands for trust that you've grown up with or, yeah or whatever all they are.
00:36:49
Speaker
ah showing the consumer that they're doing everything to to to ensure that you should trust them and they're that the voice that they're triple checking facts and all this sort of stuff before anything goes into the model. So I'd like to think that there is an opportunity for a return to trust when, you like you said, people start just going, I can't believe anything anymore. So where do I turn to try and believe? And hopefully they turn back to premium publishers and if they're smart enough ah They can take that opportunity. I guess the, you know, can coming as an American, ah you know, in the States, you've seen massive amounts of media move off those platforms, though, because people think they're biased.
00:37:27
Speaker
And so there people are getting and then there was a really interesting article about the rise of the far right in the UK driven mostly by Facebook groups. And those people aren't trusting yeah traditional media. And so that's where we start to get real challenges, because I agree with you.
Trust in Media and Brand Credibility
00:37:43
Speaker
I if if ah news is out today that somebody's died, I want to go to The Times or CNN and Fox and see all of that. But I'm not the average person on that.
00:37:54
Speaker
Yeah. Well, I don't want to go into the depths of the the US s and the problems there. We're far right. there but We're not going to solve that. That is a different podcast altogether. And certainly someone from Commerce Media Matters has no equity in talking about all right the state of the US and and its politics. um But yeah, look, i but that's an exhibit A, right? like yeah I mean, everybody has their brands that they trust and they've grown up with. i mean um ah To be honest, my sons might be YouTube. but yeah um But maybe there's a premium YouTube um in the future where he might trust that one.
00:38:27
Speaker
But you like it's case in point. You're now already going, if there's something that happened, I'm going to go to that trusted source to check it. yeah The opportunity, like I mentioned before, for was it which one did you mention? The Times. Does the Times try and disrupt themselves?
00:38:43
Speaker
By when you go to the Times, you conversate with it like it's OpenAI to find out information into louise kosta and have a two-way conversation that you can trust because that's your trusted yeah yeah place to go.
00:38:55
Speaker
Or you know are you maybe you can ask ChatGPT for the Times. you know mean There might be an integration there as well, right? like ask me Absolutely. and All right, Paul, we've asked a lot of questions. I think we've only got about five minutes left.
00:39:08
Speaker
Is there anything you want to ask us or I know you'd like to... raloit we we On our podcast, we normally wrap up with a really quick fire round around commerce. um So ah i'll I'll start on, it's just literally word association. So okay I say a word and we'll start with you and you say what pops into your head.
00:39:27
Speaker
Okay. AI. Everywhere. Credit card. Debt.
00:39:33
Speaker
Influencer marketing. ah Content creator. South by Southwest. Austin.
00:39:43
Speaker
Commerce Media Matters. Great podcast. I think that's a great way to end it.
00:39:50
Speaker
So good. Excellent. Well, Paul, thanks for joining us here at South Park. What a collab, guys. Thank you for this grab rating. So I hope people found that interesting and and we'll ahll we'll get it up online very shortly and see ah how many people listen.
00:40:04
Speaker
Fantastic. Can't wait. Thanks, guys. Thanks, guys. Take us out. Thanks, to everyone. And maybe see you next year. Maybe see Nicky. I've got to do my outro, don't You've got to do your outro. Okay, sorry. It's your thing. Yeah, this is our big thing. Okay, hold on. Wait, wait, wait. Do you want me to go for it?
00:40:18
Speaker
Can we shake it Do you want to do it? i i i I will, but I don't have the neat radio voice, so I'll do the best I can. And remember, the intelligence may be artificial, but all the wins are real.
00:40:29
Speaker
Ooh. Nice. Nice way to finish. All right. Thanks, guys. Thank you. Thanks, Paul.