← Next in Media
Next in Media
Is AI Going to Ruin - or Bolster - the Creator Economy?
00:25:47
Next in Creator Media spoke with Paul Greenberg, CEO of Butterworks, on how his company uses AI to help brands make more successful social video content, and why so far, the technology has been a net positive. Still, Greenberg talked about the dangers of the proliferation of AI slop and why it's going to become challenging for consumers and brands to sort through what's real, what's not, and what kind of attention is most valuable.
Loading summary
Transcript52 lines
- [00:00]
View Planner
Looking to get more out of your YouTube advertising? With view Planner, you can buy with total confidence and clarity. View Planner is a verified YouTube brand suitability and contextual targeting partner in the YouTube measurement program. Recognized by Google as a trusted and independent leader in driving and measuring performance on YouTube, their advanced technology gives you a clear edge. Only with View Planner we're talking custom curated contextual collections, exclusive content strategies, transparent reporting, and AI driven optimization to deliver real results. Don't let your budget go to waste. Partner with View Planner. Experience the difference. Visit viewplanner.com now to unlock your full advertising potential on YouTube. That's V U E P L-A-N-N-E-R.com this.
- [00:40]
Mike Shields
Week on Next in Creator Media, I spoke with Paul Greenberg. He's the founder and CEO of Butterworks, an AI production and consulting company for brands and media organizations. Paul was previously an exec behind College Humor, so he's really watched the creator space evolve dramatically over the past decade plus and what's interesting is Paul is seeing two seemingly conflicting trends emerge, creators and viewers gravitating to more long form episodic content while at the same time taking to synthetic creators and AI knockoffs. How this plays out may determine the future of this industry and Paul has lots of thoughts there. So let's get started. Hi everybody. Welcome to Next in Creator Media. I'm Mike Shields. I'm here with Paul Greenberg. He is the CEO of Butterworks. Hey Paul, thanks for being here.
- [01:30]
Paul Greenberg
Thank you for having me, Mike. I appreciate it.
- [01:32]
Mike Shields
Excited to talk to you. I've known you for a long time in a lot of different roles in this industry. I want to ask you about the creator space and what brands are doing there and when AI and all these cool things. But for the folks that don't know what's Butterworks, it's evolved quite a bit, I'm sure, given what's going on in the industry. What is it and where was it? Where did it start and where is it today?
- [01:52]
Paul Greenberg
Well, let me give you a little bit more of my background. Maybe it's helpful because I've been in media and digital media my whole career. I was CEO of College Humor. I was CEO at Nylon, I was at A and E Networks where I ran their digital video studio and one of their cable networks. I ran digital at Time Inc. And a TV Guide. A hundred years ago I was at mtv, which was an awesome experience, particularly back then, but one of the jobs.
- [02:16]
Mike Shields
In media really at one point.
- [02:18]
Paul Greenberg
Yeah, exactly, exactly. I got very lucky. But about Seven years ago, I founded Butterworks. We're a digital video company, or digital media company really, with a big focus on AI. We've been using AI since 2018 when we were founded. Started off using it on the predictive side, meaning we were trying to figure out how to help people understand what kinds of videos to make to be most successful on social. So we were using natural language processing and computer vision and things like that to say, like on TikTok, you should, you know, your video should be this long and they should focus on this and they should open with a human expression, et cetera, et cetera. And we've evolved quite a bit to do a lot of generative AI, as you might imagine. We also go in and kind of do AI101s for people. We will sort of set the stage about what's going on in the industry and then we'll pick a. A project to work on them with that'll. That'll help them advance their business. So, you know, we. And we also do straight video production, which is still a thing.
- [03:17]
Mike Shields
So. Okay, I want to. I was going to ask you, because 2018, you're like really early talking AI, and I don't even. I wonder if how many of your customers were ready to think that way versus, like, were companies coming to you saying, I just want to figure out how to get stuff on Snap and TikTok and whatever? Or is it always. Was it always the premise behind what you're doing?
- [03:37]
Paul Greenberg
No, it really evolved, you're right, to a large extent. Where, where people were saying, hey, will you make me, you know, will you run my Snapchat channel? For example? And we did that. But we also were starting to use AI at the very beginning, like I said, to help them figure out what kinds of videos they should make to be most successful. So if you are Animal Planet, for example, we can come to you and say, you know, don't focus on dogs and cats. Those are, you know, those videos are. It's oversaturated. There are too many of those out there. Focus on animal babies and animals giving birth. There's a much higher chance of success because there are fewer of them out there, but the ones that are out there do really well. And we were using, you know, proprietary AI and clustering techniques and things like that to figure out what topics and subtitles would be most successful.
- [04:24]
Mike Shields
I think the reason I asked about that is because I think people, maybe it's just me, the people associate either AI is either letting the machine do everything and you lose Control or do all the credit. This is like a different way of using it.
- [04:36]
Paul Greenberg
Exactly. That's exactly right. So we would use computer vision to look at the video, you know, this whole tranche of thousands of videos about animals to figure out what was working and what was going on in the videos. And you know, and then we would match it up with metrics to figure out what kinds of videos were most successful and how long they were, like I said. And as well as, you know, trying to understand, should you start your video with an angry expression or a disgusted expression or a happy expression? And turns out happy expressions don't do that. Well, when you start off with that, you want to give people a sense of intrigue and shock. So you really want to start your video out with a disgusted or an angry expression in some cases. And we weren't replacing the creatives to be very clear. We were giving the creatives lane lines to play in. So again, it's like come up with a generic animal video, you're totally blue sky. It's really hard to do that. Or come up with an animal baby animal giving birth video. And that's going to give you, we.
- [05:31]
Mike Shields
Talked a lot about animals giving birth so far.
- [05:33]
Paul Greenberg
Sorry. Yeah, but it's going to give you guardrails to play in and just make it easier for creatives to do their jobs.
- [05:40]
Mike Shields
And how, how were you doing this then? Because you were, you didn't have chat, GTP and all these things to tap into. Were you just like, you know, doing your own analytics? Like, were you pulling from a whole bunch, bunch of different APIs. What was this? What was this like?
- [05:53]
Paul Greenberg
Yeah, we were pulling from APIs from the platforms and we, we wrote a lot of our own code. We had a data scientist on the team who was doing, you know, all a lot of that work. We were using AWS to do a lot of the computer vision and the natural language processing work. I mean, obviously we weren't going to be able to build that ourselves, but we didn't need to. So there were, there were a couple of off the shelf solutions that we looked at, but there was a lot of proprietary code that we ran to do this and to figure out those topics and subtopic clusters.
- [06:22]
Mike Shields
Okay, so I do want to come back to the use of AI especially fast forward to where you are today, which is going to be, I think, really interesting. But you've watched this evolve over time where they, where brands and agencies would subcontract this kind of work, would find someone who helps them get going on snap to or figure out how to work with creators. How has it changed overall? Like what kind of trends are you seeing in this creator space when it comes to media advertising and are we doing enough there generally speaking?
- [06:50]
Paul Greenberg
You know, it's interesting. I think that kind of what's old is new again in the sense that videos are getting longer. So all the platforms have increased their maximum time. You know, TikTok, you can now do a 60 minute video. YouTube shorts went from 15 seconds to three minutes on YouTube proper. If you're verified, you could basically make it as long as you want. So you know, I think that we're seeing, I don't want to say a return to traditional television in that respect, but we are seeing longer videos that are creating engagement that people are expecting to, to spend more time with. And I think that's only good for brands because it gives you more of a chance for product placement. It gives you more of a chance for pre, mid and post roll or you know, commercials if you want to call them that. But I think that it really gives people the opportunity on the brand and the agency and the media side to interact with an audience in a, in a way that's comfortable and extensive in the sense that it gives them lots of choices and lots of opportunity.
- [07:49]
Mike Shields
Why do you think that happens? Is it. I mean, it's probably hard to answer that across the board but because, you know, I think the presumption was everyone's attention span was getting worse. Everything was fee driven and random. But then you had creators who were having these obviously building up real audiences and looking to do bigger projects and consistency. Like was it a business move by the platform so just you can squeeze in more ads with longer stuff. Did the consumers lead us here? Like why is this happening?
- [08:13]
Paul Greenberg
I think it was both. I think it was push and pull in that respect. I think that the platform said, you know, hey, you know, we're not seeing enough ad revenue and we want to give people more opportunity to monetize. But I think on the audience side, I think there's a place for super short form video and you know, TikTok reels, YouTube shorts still offer that in spades. There's lots of, you know, very small, you know, very short videos on those platforms, of course. But I think that the, you know, consumers wanted to go deeper with their favorite creators. You know, if you're a markiplier fan, you want to know more about him, you want to sort of see more of what he's doing. Obviously Mr. Beast, you know, you know And Kai Senate and all those people are doing a lot with much longer videos. I mean, Kaisen video was a month long. So, yes, different. Different beasts altogether. But, you know, I think that the audiences of creators want to get closer to them and want to know more about them and want a connection to them in the same way that old TV audiences wanted a connection to their favorite characters.
- [09:14]
Mike Shields
So when it comes to explaining that to brands, when you work with them, are they make something that the algorithm is going to love versus making something for viewers that people are going to love when they're on there and their creators should know that. How do you do you get those questions and you struggle with those that balance your clients?
- [09:32]
Paul Greenberg
We do. We struggle with that all the time. And, you know, obviously you want to go viral. You want as much as many views as possible. You want as many new followers as possible, you want as much engagement as possible. Some of the work I was describing before about, you know, the insights that Butterworks was doing helped with that, where we were able to say, you, you know, this is a little bit of art and science mixed together where, you know, we can tell you again, the guard rails you might want to play in. But then you got to bring your own creativity and your own special sauce to the party. But I think that, you know, it. It really gets to a point where the algorithm is almost impossible, not impossible to figure out, but it's very difficult. You know, I mean, right now on YouTube, what's fascinating is the algorithm isn't working for the platform. It's working for each individual audience member. So you and I have different YouTube algorithms that it feed us in terms of recommendations. And, you know, I mean, they've been talking about this publicly recently where they're saying, you know, watch time matters and thumbnails matter and ab testing of, of how titles matters. But at the end of the day, you want to figure out how, how you can have the most number of people, the biggest number of people, really gravitate towards your content. But again, you're. You're getting recommendations that I'm not getting.
- [10:49]
Mike Shields
So that's change. It makes it difficult because you see creators, and this has been talked about a lot, they, for the same kind of thumbnails, the same kind of headlines, because they think they figured out something in the algorithm. Is that, is that changing now because it's, it's more individual, or has it always kind of been that challenge?
- [11:05]
Paul Greenberg
I think it's changing. You know, back then there were cruder tools for targeting. There was, you know, Collaborative filtering. So it was like the, you know, putting people into cohorts and not being able necessarily to say, you know, Mike is going to like this kind of video because he's watched know these other videos. And so I think it is changing and I think it's going to make it more difficult for creators to continue to build audiences because they don't understand and they can't possibly understand the algorithm for each individual person. Which means that you've got to just stay true to what you're building and try new things, but have your own authentic voice come through. And, and this has always been a problem. You know, if you're trying to fix the machine and figure out how you can create content and media for an algorithm, it's going to be very difficult to make it work just on that one to one basis. You've really got to bring some soul and some creativity to the table as well.
- [12:01]
Mike Shields
Okay, I know we're talking very broadly here a lot besides animal births. Give us some stuff that. What are some kinds of things you're working on with brands that are really working right now? Either, you know, really interesting campaigns or collaborations with creators that people pay attention to?
- [12:17]
Paul Greenberg
That's a good question. We can't talk about too many of them. There are still a lot that are private, but we're working with a number of different parties. For example, this is kind of interesting, we're doing generative AI in certain cases where we're actually using tools to create video out of prompts or out of pre existing images or videos. That's still good, but it's still not. You'd still need to do a lot in post production to make those kinds of things work. But for example, we're working with a healthcare network of cardiologists who are building a video portal for patient education. Your provider gives you a QR code, you scan it, you watch a playlist. But we're using AI to create digital avatars of the doctors to create customized videos for patient groups. So your doctor introduces a playlist of videos that you can watch. It makes it more customizable, it makes it, they're better outcomes for patients. It causes fewer phone calls to the doctor's offices and it gives brands opportunities like device companies and pharma. It gives them opportunity to, you know, really get in and, and work with specific cohorts of patients and doctors that they really didn't have the opportunity to do before. So that's, that's sort of one of the projects that we've been working on in the past, it's been really interesting. We're working with a couple of other really interesting creators, but we're really. We don't work generally with individual, individual creators. We mostly work with larger companies and brands. But I think that another one of the trends that's been interesting, everybody comes and says, okay, use AI and eliminate all the people.
- [13:56]
Mike Shields
My next question, where are you now with this? Because my mind goes through the negative place always. Exactly what do we need Paul anymore for? We just plug it into. Right?
- [14:06]
Paul Greenberg
We plug it into a machine. What do we need Paul as the executive for and what do we need Paul the creator for? If I have a successful YouTube channel and we're not there yet, and I don't know that we're ever going to be there specifically, but you look at, like, Lil Michaela and you look at, you know, AI generated creators who are incredibly successful. Although it's interesting with little Michaela and people like that, their engagement isn't quite as high as it is with human creators. And that may be that people, you know, gravitate towards what she's wearing or what she's, you know, touting in terms of outfits or, you know, other things. She talks about music, but there's still some gap.
- [14:48]
Mike Shields
But the relationship is not the same.
- [14:50]
Paul Greenberg
It's not quite the same. Exactly. That's the data that I've seen. And again, I'm sure your mileage may vary, but I think that there are a lot of people who still do want that human connection. And I also think that it's incumbent on creators to use AI to augment their work and to be creative. So there are people like Matt Parr, who has 12 different channels that are all faceless videos where it's just BO over stock footage. But he's, you know, but he's using AI to create all of that. And, you know, it's doing very well. There's like Chase Reiner and Isabella Cotsias, if you know them, they're. They're actually using it, not necessarily to create the video up front first, but they're using it to figure out trends and sort of like what I was describing with what bundleworks was doing before. So I think that there's. There's a lot of different ways you can use AI to augment the kind of work that you're doing. But it's still, you know, you still do need, you know, humans in the loop. Having said that, there, you know, I think another interesting way to look at this is deep fakes. And I don't know if you saw this, but ByteDance just announced, they haven't released it, but they announced something called Omni Human, which is one of the most convincing deepfake technologies that I've seen so far. You know, they deep fake Taylor Swift singing at a concert and all kinds of other crazy things which you've seen before. This seems to be technologically a step ahead.
- [16:16]
Mike Shields
A little bit better than that time.
- [16:18]
Paul Greenberg
Exactly. It started so, so we're getting there too. You know, as I said, sort of with the doctors, where we're using digital avatars to have doctors, quote, unquote speak, where we had them train a video for three minutes and then use text prompts to just have them talk for as many hours as we need. There's going to be a lot more of that as well.
- [16:38]
Mike Shields
A little bit of random question, but with, with the open, there's a lot of fear on the open web and I guess it's already really happening where you're going to see a flood of AI slop because that is what is going to the, the new search engines you're going to respond to. You know, it sounds like you're talking about with there are these like synthetic creators on YouTube and TikTok and there are these creators using this, these tools to enhance what they do. Are you seeing, are you worried about these platforms being flooded with just total garbage that is meant to populate search results, take monetization from somebody else? Are the platforms on top of it? Like, should we be freaking out about that or I don't know, for free.
- [17:15]
Paul Greenberg
We should freak out about it. But it's absolutely happening. And you know, Google's admitted they're having trouble keeping up with all of the new AI created content. But then again, you know, they're sort of perpetuating that in a way. Where they've got the AI Gemini overview at the top of many searches saying no to it.
- [17:34]
Mike Shields
Right? It's right.
- [17:35]
Paul Greenberg
Exactly. And how do you, how do you SEO for that? It's almost impossible because there's sort of one result that they're, that they're giving. And before they used to, they still had those with their overviews where they give one result, but then they had links closer to the top where you could choose and kind of scroll down and see what you were looking at. These are really being presented in a way that's authoritative and obviously it's made a lot of mistakes. You know, glue on pizza and other kinds of things that, that we've seen. But I think that, you know, it's going to become harder and harder to separate the wheat from the chaff with those kinds of things and figure out, you know, what's real and what's going to be worth my time. You know, again, it comes down to the attention economy. At the end of the day, right? There's a finite amount of attention that's not getting any bigger. And so at a certain point, as a consumer, am I just going to turn it off and say, you know, I can't tell what's good and what's bad. I don't know what I'm going to be fed, so I'm going to go somewhere else and try to figure out how I can do that. And so, you know, I think that there's a lot of, of those kinds of things that are, that could start to happen as slop proliferates coming back.
- [18:44]
Mike Shields
To your company and how you work with, with brands and media agencies and have things gotten any cleaner or smoother in terms of like where the budgets come from and where the decision making. Are you mostly getting production work through social video budgets? Are you working with media budgets overall? Is it like, is it all over the map still in terms of how, how it happens?
- [19:06]
Paul Greenberg
It is all over the map and it depends on the company. And even within the same company, we're getting budgets from different places. So sometimes it's the research group, sometimes it's the marketing group, sometimes it's the content group, which, you know, is all good, but I think that it sort.
- [19:21]
Mike Shields
Of takes the money.
- [19:23]
Paul Greenberg
Exactly. But it speaks to the fact that I think, you know, media companies are still trying to figure this out. And look, a couple of them have done some interesting experiments, right? I mean, Runway did a deal with Lionsgate to have them try to, you know, train on their corpus and help them create new content. Blumhouse did a deal with Meta. And then, you know, within the companies, obviously Netflix has been using AI forever and they're doing a great job with it. Disney has a product called Face Director which allows composite takes of facial expressions that they can put together and combine in an interesting way and then the de aging kind of thing. So you know, Adobe has product called Sensei which is great for post production, removing objects, things like that. So there's, there's a lot of things that media companies are doing as they kind of go down this rabbit hole and try to figure out how to use AI. But then again, you hear, you know, the Brutalists editor said, oh yeah, we used AI a little bit on there. You know, we Use respeacher on the Hungarian accents for the characters and huge amounts of blowback. So it's still very muddy in terms of what's okay and what's not. I feel like if you use AI as a tool and people get paid for it, then there's no, you know, then. Then that's okay in. In a lot of ways. So if you have an actor who allows you to use her. Her likeness for a movie, but she didn't have to act in it, but she's still getting paid, you know, I think that that's even better. Right. You'd have to get out of that.
- [20:52]
Mike Shields
I guess the question is we had to just how much does the audience deserve or want to know about these things or care? And I think that's probably an open question.
- [20:59]
Paul Greenberg
Yeah, that's a great question. That's a great question.
- [21:01]
Mike Shields
All right. Totally unrelated, but very of the moment, I'm asking. I'm asking everybody this, like, I don't know what's gonna happen with TikTok. It's very confusing. Can anything replace it? I've heard very different opinions that everyone will just go to the Reels and you in Jeep Shorts. It's such a different audience and different algorithm that that won't happen. It's up for grabs. Do you have any feeling sense of what might happen if this ban actually takes place?
- [21:25]
Paul Greenberg
I mean, it was fascinating to see Everybody go to RedNote all of a sudden, right out of nowhere. And I think that, you know, RedNote has a very similar algorithm to TikTok, which I think was what was so appealing. But I think that the Reels algorithm and the Shorts algorithms are very different from TikTok. So I don't know what replaces it, to be honest. I think people will gravitate towards those two platforms because they want to consume that kind of content. I think that obviously Instagram particularly is trying to adjust to be more like TikTok with rectangular thumbnails and things like that, but I think that it's going to just simply be a different kind of experience. I don't know that you can replicate what TikTok is. People still will crave that type of content interaction. They're still going to want to see their favorite creators and their favorite cat videos and things like that. So I think that people will gravitate towards that. Those platforms are growing anyway, and, you know, they do a great job of giving people what they want.
- [22:27]
Mike Shields
I think you'll see a YouTube or reels get very aggressive and try. And they're Already they're already trying to snag creators and they seem to want to push those apps in front of people more. It's very unpredictable that you can't just assume that the crowd that's on TikTok regularly, and I know it's very broad, they're trendy by nature because they're younger. So it's just really hard to predict. I was going to ask you, do you think people joke about the. That's like zombie time or brain rot and it's not great for you, but they'll, but they get stuck in it. Do brands think about that like when they're mindless scrolling or is it seen as like super engaging and really valuable?
- [23:02]
Paul Greenberg
I think again, it goes back to attention and I think they want to be associated wherever there are people's eyeballs. And what we're seeing is people are trying to figure out they would love to be part of those audiences. So I don't think they're saying, oh, this is brain rot, we don't want anything to do with it. I think if, you know, there's types of content that are objectionable, obviously, but generally I don't think they have. There's a sort of ethos that says, oh, we're not interested in TikTok because it's short form and because it's sort of mindless scrolling. I think the problem everybody's having is trying to figure out how to appear authentic as a brand. And so you have, you know, a couple of breakthroughs like Duolingo, like Arby's and people like that who are doing great jobs of figuring out how to really sort of insert themselves into the zeitgeist. But I don't, I don't see anybody saying I don't want to be part of that because it's, it's not good enough for us. I think what they're saying is how do we be good enough for it?
- [24:00]
Mike Shields
Last question is, have you seen any shift? You know, there, there's definitely. There was a hyper amount of attention and concern around brand safety and brand and suitability on these platforms in recent years. Now it's almost. We're like in this new era where the, you know, meta's like said, forget it, I'm not going to monitor anything. We have the big masculine energy trump time, like no one cares anymore. It feels like that. Are you seeing any sort of pullback on brands being concerned about these things or. Not necessarily.
- [24:28]
Paul Greenberg
I haven't, I haven't, I haven't. And I think that that you know, Meta even said last week they're not seeing any slowdown in advertisers since they announced their policy to remove the fact checkers. So I feel like brands are sort of adapting to the new world. You know, brands are really an indicator of the collective consciousness and this. And so, you know, they're, they're lagging indicators to some extent of what's. What's going on out there. And so if people are still going to be on Meta's platforms and obviously, you know, you saw their numbers, their, their daus just keep going up.
- [25:01]
Mike Shields
Half their business. Instagram right now.
- [25:03]
Paul Greenberg
Right, Exactly. It's crazy, you know, and, and, you know, kudos to them for, for being able to. To build that business like they've been. Like they have been. But I don't think. I don't. I see brands going where the audiences are and the audiences are still there and not concerned with those issues yet. As soon as the audiences become concerned, I think the brands will become concerned, sir.
- [25:24]
Mike Shields
All right, well, Paul, so many other things we could talk about, but we're gonna. Let's leave it there. Thanks so much for an awesome conversation and let's do this again.
- [25:29]
Paul Greenberg
Definitely. Thanks for having me, Mike. I appreciate it.
- [25:32]
Mike Shields
Thanks again to my guest this week, Butter Works, Paul Greenberg, and my partners at View Planner. If you like this week's episode, please take a moment to rate and leave a review. We have lots more to bring you, so please hit that subscribe button. We'll see you next time for more on what's next in Creator Media. Thanks for listening.