How much should AI decide what you watch on YouTube and determine what people like me create? Let's talk about it with YouTube CEO Neal Mohan right after this.
Welcome to Big Technology Podcast, a show for coolheaded and nuanced conversation of the tech world and beyond. We have a great show for you today. We're going to talk all about AI's dramatic influence on what we watch on YouTube and what we create on YouTube, whether it's advisable, and a bunch of new products and features that YouTube is releasing today. And we're joined by the CEO of YouTube, Neal Mohan. Neal, great to see you.
Good to see you, Alex. Great to be here.
So, I went back in the archives and I tweeted about your move from Google's ad division to YouTube 11 years ago in 2015 and I said, "Neal Mohan is off to YouTube ready to make them some money." Because he had been on the ad side. It worked. YouTube today, for those who are catching up, is a $60 billion annual business. You can correct me if I'm wrong here, but that Netflix is 45 billion. So dramatically bigger than Netflix. More than 2 billion users globally. It's growing 13% a year. It accounts for 14% of all US streaming or viewing.
And that means a billion daily hours of YouTube are watched on TV alone. And the company's paid out $100 billion to creators over the past four years. So we talk about YouTube, it's not just the cat on the skateboard or the dog on the skateboard anymore. This is a media force.
Yeah. You know, the first thing that comes to mind when you mention that tweet is time flies. It's been a long time. And you know, YouTube is really kind of everything that you described it to be there. You know, we've been around, as you know, we celebrated our 20th anniversary last year. And you're right, we're this platform that started as a place where you would upload videos like the ones that you described. And now we are a streaming service where two billion plus people come to every single day connecting with millions and millions of creators. And in an essence, that's the magic of YouTube. And as a result of all of that fandom that manifests on YouTube every day, that's what creates the business opportunities of course that you talked about for our creators, for our partners, for the brands that work with YouTube, but also I think probably even more importantly creates sort of the cultural phenomenon that is YouTube. And so that's super exciting.
Yeah. And when you think about like the ability of companies like Netflix to grow or Spotify to grow, Amazon Video, you know, when I'm asked, okay, who's the big competition they're coming up against? It's often YouTube. And, you know, you could say it's a Netflix versus HBO battle, but it's really Netflix and HBO versus YouTube. It's Spotify and Apple podcasts, which we have big listenership on, but it's versus YouTube, where I would say most of our listenership is on. And that is a big shift. But there's sort of a difference between YouTube and these other, you know, legacy media platforms, which is that you would have a group of people programming, thinking what the audience wants and then adjusting based off of how the shows do for like a Netflix or an HBO. With YouTube, it's largely people programmed. And by that I mean it's like people like me and other creators putting videos out there seeing the distribution that they get, finding an audience, seeing what the audience wants, and often times doubling down on what the audience wants. But a lot of that has been, you know, almost an inexact science for creators on YouTube. You sort of listen to the algorithm and you make some of your own decisions and then you program. But now what we're seeing with the fact that I would say YouTube is the Google product that uses AI the most is that AI is coming into the product, coming into the programming decisions. So for instance we're going to talk about some products and features that you're announcing today. One of them is drafts where a creator can now upload a draft on YouTube and get feedback and suggested edits that's coming in from YouTube's AI. When you do that, YouTube will also suggest thumbnails. And it will give you feedback. Hey, you know, your thumbnail might do better if you, you know, maybe put two faces in or look more like, you know, Mr. Beast when he's like, you know, into the camera. There's also going to be dynamic thumbnails. This is all new, which are going to be optimized based on the audience. So maybe one audience responds to one thumbnail and then the other audience responds to another and YouTube will dynamically show them what they want. You could also, this is a crazy one, you can actually AB test not just titles and thumbnails, but you can AB test videos, which is that creators can now put up like a couple of videos and maybe one with maybe we'll do that with this one where we'll do one with like a trailer, one without a trailer and see what does better, see what gets more watch time and is enhanced by the algorithm. Okay, I promise I'm getting to a question. How much do we want to let AI program what we watch on YouTube? Because this is AI effectively making programming decisions as opposed to the creators listening and intuiting versus have those programming decisions be made by humans.
Yeah, it's really an interesting way to frame the question, Alex. And I think what I would say kind of at the big picture level is we do use AI a lot. Most of the people that work for me are software engineers and our job as I always describe is to build the world's best stage, but it's our amazing creators, including yourself and all the success you've had around your channels on YouTube, that are the people that are on that stage. And that's sort of the mental model that I would ask people to sort of hold in your heads. Like if you walk the halls of YouTube every single day, you're going to hear that term creator or podcaster or artist creators at large a hundred times a day. And the way that I think about AI is in that context, which is how can we use the latest and greatest technology, you know, we work with companies like Google DeepMind every single day to make that stage even better. And so it's really about doing what YouTube does, helping creators build an audience and then build a business, but to do it more efficiently and do it faster. And so a lot of those features and products that you just described, including some of them kind of hot off the press that we're going to talk about at our Made on YouTube event, are powered by AI, but they're doing things that are core to the creator business. So, dynamic thumbnails, why not actually give creators the ability to have multiple thumbnails associated with their video depending on who the audience is? AB testing, right? That's a concept that's existed for a while. Why not put allow AI to put that in the hands of every single creator? And those are all powered by our investment in AI using Gemini, our creation tools like actually being able to hit the plus button on YouTube and creating videos that's powered by models like Veo, but it's all in the service of human creativity like what creators are looking to do around their creative ideas but then also around their business ideas and that's the mental model that you should have going back to sort of the core AI use in YouTube, which is when you open up the app on the television screen, those recommendations that you get, the reason why they are so relevant and useful to you as a viewer is because of our investment in AI. And so that's really the way that I think you should think about our approach to the technology as a tool to empower human creativity and to bring you know those billions of fans that I described much more closely to their favorite creators.
Yeah. And there's by the way I don't think any platform can do what YouTube is doing just because of the level of AI you have access to. So it's a real advantage over the competitors that I mentioned earlier. But then I just want to go back to this notion of like how much do you want to let the AI decide? Because if you think about it, like this is the arts in a way. This is where the arts are shifting. And you know, there's always been an element of like, you know, just kind of throw something out there and see and you know, maybe it doesn't appeal to everybody, but you throw it out there and, you know, over time that is what becomes the art form. Creators, because I know I'm on YouTube, they'll respond to incentives. So like if the algorithm, if YouTube even today suggests videos to make because there's an audience for it. So, you know, they'll respond if they see that there's like an audience to be found, you know, by creating those type of videos. Or if they see that if they follow the AI suggestions on where to edit, they'll do it because they think they're going to do better. And so, I guess like the big question is how much like is it a good idea to be trusting the AI to make all these programming decisions and hand the art effectively from our hands to its hands.
Yeah. You know, the way that I would answer that question is frame it a bit differently. So implicit in what you said is we all as the ones who are sort of coming up with the creative idea or picking and choosing the right type of content, we know better. We get to decide what to put in front of the audience and that's the way you know traditional media and Hollywood and all that has worked for decades and it's been amazing. Lots of great art has come out of that and from the very beginning YouTube's approach has been very different. Our mission is to give everyone a voice and show them the world. And that the first part of that mission means something very specific which means that there are no gatekeepers. Nobody's saying that, "Hey, Alex, you know, that idea is kind of cool, but maybe you shouldn't try it, or maybe you should do it this way." You get to decide as a creator, this is how I want my podcast to be. This is what I want it to stand for. This is what Big Technology is going to represent and fill this niche in the industry, and it's your choice. So, you're deciding to do that. And what AI does, and you know, I often say replace that word AI algorithm with audience. It's the audience that is choosing. That's the difference between YouTube and other platforms. So, it's not about the AI or the algorithm. Like I said, that's the stage. But what you're seeing in YouTube every single day is the audience giving you feedback. And so, why should you know three or four people sitting in Hollywood make all those decisions for all of us as viewers? Why can't the two billion viewers who come to YouTube every single day make decisions for themselves about what they like, what they don't like? And it's that sort of entrepreneurial sort of creative energy that exists on YouTube every day that has allowed millions and millions of creators like yourself to flourish. And so it's not about the AI picking the content. It's about us as viewers picking the content, the exact content that we want. And nobody gets to tell us on YouTube what's good, what's not good, what's art, what's not art. We get to decide. And that's the power of an open platform like YouTube.
Yeah. And I should say I'm much happier with the YouTube version of the world than the gatekeeper version of the world because let's face it, if I was trying to do Big Technology in the world of gatekeepers, it probably wouldn't happen. Like we exist because there are no gatekeepers. We don't need to go to the New York Times to publish articles about technology. I don't need to go to HBO to be able to make this show.
Yeah. And you said something very interesting as a creator which I think a creator really understands intuitively which is how to make tweaks, how to respond to feedback. You know, you described it in the context of the algorithm. Well, that's just the audience giving you that signal, right? The other place where you get feedback instantaneously is right below your video on the watch page. Those comments, right? Like they are telling you like whether you should do audio versus video or what have you and that feedback is instantaneous and that is literally your fan base, your audience, your listeners, your viewers telling you exactly what they think about what you just did. And then it's your choice then how to incorporate it or not incorporate it as opposed to having like 10 middlemen in between deciding that for you or that feedback coming 3 months later after the you know quote unquote season is done and it's too late.
Yeah. And they're not shy. So, be nice, folks. But let's talk a little bit more about this because, you know, these are all the good things. The thing that I think and it seems pretty clear the way that the world is heading, this is going to be like the way that all content goes. Maybe not all, but much of it like YouTube has hit shows on it. YouTube is this is another announcement you're making. You're leaning into more shorter miniseries on YouTube. So, YouTube is starting to sort of it's not even starting. it is, you know, well, beating the legacy players, but in league with that level of production as well. The thing, so as we look to this, I think it's good to talk about the thing that maybe I've been getting at and haven't been clear on. And my questioning is, you know, doesn't it all kind of can it all end up looking the same? Like, have you heard of like I think it's called Instagram face. It's like the picture, the way the same pose that everybody makes when they look at the camera and they put it on Instagram. Like there is an algorithm-driven homogenization of content that's happened there. And so how do you think about that and how do you guard against that?
Yeah, you know, I think that's just not the way YouTube has worked. You know, you've obviously followed the YouTube story for almost since the very beginning. I have as well. And I think that it honestly ends up being almost the complete opposite of that. And what I mean by that is what happens because of this lack of gatekeepers because of this sort of entrepreneurial sort of creativity that YouTube allows like you describe in your own case. There are millions and millions of fandoms and niches that emerge. You know what I think the YouTube experience is a bit of the opposite of that, which is when you turn on the TV, you get very specific episodes or shows or creators based on your interests. Now, it turns out that there are millions of other people who also have those interests. So, those niche sort of cultural phenomenon are not tiny. They're large because of, you know, the two billion people that come to YouTube every day, but they really are millions and millions of these types of niches. And the same thing happens to me in my experience. And by the way, it changes through the course of the year. And I happen, you know, I go to YouTube because I love to learn, you know, kind of random esoteric things sometimes, but I also go to YouTube because I love sports. And that's sort of as mainstream monoculture as it gets. And sort of YouTube sort of gives you that sort of full spectrum as opposed to any sort of homogenization. What I would say is that because we are, you know, used so broadly really all over the world, it's very hard for that kind of homogenization to happen. And it's more about sort of these very kind of these niche interests that are also followed by, you know, a million of your closest friends, right?
Well, there's a degree of that. Maybe not in the videos themselves, but like if you think about like it's kind of this phenomenon on YouTube that the way that you know Mr. Beast sets a lot of these trends and then people follow, right? Like the way that Mr. Beast looks at the camera in his thumbnails is something that if you once you take a look at it, are you doing those in your thumbnails?
I'm going to pay closer attention. You know, we found that like the shocked expression doesn't do as well as just kind of like looking concerned or a laugh or something like that. So, but we're yeah, I don't know. I don't think we're in the same category, but there has been some of this like I would call it maybe algorithmic driven homogenization in some elements on YouTube. You don't think so?
You know, I think that certainly there are creators, up and coming creators that look to, you know, Jimmy, just given his success on YouTube and just his success in the culture as, hey, that's a blueprint that I should try to follow. And look, and by the way, I think that has actually always been part of the YouTube culture, like the YouTube creator culture, to a degree. No, like the aspect of if you've made it, your job is to actually help the next generation of creators come along. And I think creators like Mr. Beast and others really lean into that. But there is that aspect of trying to learn from those creators that have been successful. What I'm commenting on though is this notion of sort of kind of homogeneity across all of YouTube. And I certainly we don't see that in the data. And we certainly don't see every time I go, I always, you know, one of my favorite questions is when I'm at an event or, you know, hanging out with friends is just to ask what their favorite thing is to watch on YouTube. And I can tell you I've been asking that question for, you know, a decade now and I almost always get a completely unique different answer. And I think that is more the essence of the experience on YouTube and frankly where the longevity comes from.
Okay. Yeah, it's interesting. I mean, the AI might be helping, you know, more with, let's say, packaging when you think about this, the dynamic test for thumbnails and maybe there's already AB testing for headlines. So, okay, it's good to know it's not going to.
I mean, it helps with all of those things, but it helps with really just like I said, it's like a it gives us a different set of toolkits that help with different aspects of YouTube. You know, one of the core parts of YouTube that I know you're familiar with is Content ID and protecting creators' rights and their IP that was investment in AI well over 10 years ago. So, it's sort of deeply rooted in every aspect of the company and the product. And there are places where it's more visible, right? For example, like those creation tools that are right behind the plus button on the main app, but there's places where it's just kind of deeply embedded into how YouTube works.
Totally. All right. So, when you think about the content that you want on YouTube, do you want AI generated content on YouTube?
The way I look at it is we want content that connects with viewers and fans all over the world. I often sometimes say that, you know, YouTube is the world's most efficient connector of a creative idea or creative person with their audience, no matter where that audience is in the world. And so to the extent that that is the core of what YouTube does, I want more of that type of content, right? And whether AI is used as a tool to create that type of content, that's just more a manifestation of sort of the state-of-the-art technology that exists today. I do believe that it is YouTube's responsibility to make sure that our creators, like yourself, have access to those tools, but it's the creator's choice and responsibility to choose which tools to use and how to use them. And what I find is when viewers come to YouTube, it's really more that human story or that human connection that they're looking for. And it tends to be that that type of content is what rises to the top. And I think AI in that sense is not different than other types of technology that enabled creativity in the past, whether it was, you know, the advent of the synthesizer when we were growing up, right? Like or the drum machine or what have you. Like there's always technology that comes along that enhances the artistic experience. But what is also true is that there's a human being that actually is sort of the maestro that's controlling it. And I think that that's at least so far what we're seeing on YouTube. So we are seeing I see it in videos you probably see it too the use of these new technologies to actually create a palette of content that otherwise might not have existed but it's all in service of that human storytelling and so that's what we're seeing on YouTube both on the short side but also on the long form side because I think people come to connect with humans. You know, two computers have been able to play chess against each other for a very long time. It's really only interesting on YouTube if at least one of them is human, right?
And some of the AI videos, I mean, they're getting good. I don't know. Did you see the video recently that's been trending of the Coke truck that sprung a leak? You didn't see this one?
Neal, you got to watch this. We'll put it up when we play the video, but let me narrate it for you. There's a Coke truck on the side of the highway springs a leak. A man with a gallon jug walks up to the Coke, the Coke streaming out of the back of this truck, puts the gallon jug in the stream, and starts chugging the Coke. Then he opens up the truck next door, which happens to be a Mentos truck, and he starts dumping the Mentos out on the sidewalk, mixing it with the Coke on the floor, and it creates an explosive propellant, and he shoots out of the frame. He stands on it, and he's gone. Now, how many times did you watch it?
Many times. This was most of my weekend actually. I wouldn't say that's AI slop. I mean that's good.
Actually, you're hitting on a key point there, Alex, which is you asked at the beginning of your question like how do I think about AI generated content? Do I want more or less of it? And I think that's not the dimension along which I think. And so hopefully my answer conveyed that AI technology might be a means to that creative end, but what rises to the top is new creative ideas and concepts and things that resonate with creators' fans. And so there isn't an explicit sort of like AI bad, everything else good or vice versa. All the kind of rules of starting with our content policies on our platform to you know what resonates with audiences are independent of that factor.
So you're saying basically if it's good it's okay.
Yeah. Yeah. I think so because like I said what comes through even in the example that you described is there's some aspect of storytelling associated with there's and by the way the technology is evolving so fast that it's also very easy to create low quality content and so that's probably more the dimension into which to look which is you know whether it's referred to as like AI slop or what have you that's different than AI per se, right? Like it's always it's about high quality content.
So, YouTube won't like penalize content if it's AI off the bat. They're going to give it a shot.
Not if it's not purely on that sort of aspect of it, right? And by the way, like what constitutes AI, right? Like did you use some tool to do this 5-second thing in your video or this in the background? Well, people have been doing that since the beginning. Purely AI content.
Yeah. So, but like how do you even there, how do you draw the line, right? Like how is that put together or not? And so, like I said, it's got to be more about the quality of the content and what's resonating with an audience and having that rise to the top. And we take the notion of AI slop very seriously. I would never want, you know, our YouTube Shorts experience to basically become that or any of our experiences or surfaces on YouTube to become that.
Okay. So, you believe So, the term AI slop, you believe that there's a thing called AI slop and you don't want it.
I mean, you I mean, I think you've seen it. We I've seen it. It's all I made it, which I have, but I'm trying not to upload on YouTube. But I have like thought about like creating other channels where you know for instance I would take the day's tech news put it through Notebook LM which is a Google product and you can make videos from Notebook LM and then upload it to YouTube. I don't know if that's kind of like in the border of like slop and of interest.
I mean again I think it has to do yeah how the audience responds. Not just how the audience responds. Yes, that's important as we talked about earlier, but also how you as the creator feel about the content of that video, right? Like does it carry forward what you stand for editorially? Is it something that your audience expects from you or is it just some kind of random kind of video that's uploaded? Like those things actually matter. And I think that's what goes into what succeeds on YouTube versus not as well.
Since the development of or the inception of these AI tools, have you seen an increase in uploads on YouTube? Has the curve bended?
I don't think it I don't think necessarily directly correlated to that. I mean YouTube continues to grow, right? Continues to be a place where new creators are finding audience every day. Like I get sometimes I get asked this question. It's like well you know who are the most important you know creators on YouTube or who do you think about the most? And what I always say is the most important creator is actually that person who's hitting the upload button for the very first time today because that's the lifeblood of YouTube, which is creating an opportunity for the next Alex to come along and start uploading today. And so as YouTube has become this cultural phenomenon, of course, there are many more creators, millions of them kind of trying their hand at it. I think that that's independent of, you know, this question of AI or not. Does AI actually make some of this content creation easier and further democratized it just like YouTube did in the very early days 20 years ago? Yeah, of course. And I think that's a good thing.
Okay. So there was not like you know some level of content creation happening on YouTube and then just an inflection point with these AI. That's very interesting. Okay. One more question about this then we'll move on. Do you view AI as a competition for you? Because, you know, there was that interesting comment from like the Netflix CEO where he was like, "Well, my competition is sleep." And it's like, well, for YouTube, you can really only be watching one thing at the same time or engaged in one, maybe two, maybe. People, they'll watch something on the TV and then they'll have shorts on in their phone. But with AI conversational AI bots are going to become much more engaging, much more especially as the AI gets better, they're going to command, I believe, a bigger share of people's time. They'll be an entertainment medium. And so I'm curious to hear your thoughts on like do you view that as like when you think about when you do your big planning and you think about competitive landscape has like AI chat bots even registered for you as some place that you might have to fight for share of time with?
