Venture Deep Dives: Generative AI & The Future of Consumer Content
Find out what two things the industry must do to adapt

Our “Deep Dives” series explores vital aspects of burgeoning VC sectors with one of our ~50 investing experts. This week, Principal Grant Demeter discusses the impact of generative AI on the future of consumer content. Learn how generative AI could add as much as $4.4 trillion in value to the global economy.*

Deep Dive: Generative AI & Consumer Content
Grant Demeter, an investor and principal with Alumni Ventures, recently shared his insights on the future of consumer content driven by generative AI. Find out what two things he believes the industry must do to adapt, and hear his predictions for the future of content.
See video policy below.
Interested in Investing in Startups Like This?
Reserve Your Spot in Deep Tech Fund 4
We are seeing strong interest in our Deep Tech Fund. If interested, we recommend securing a spot promptly.
Investors in the fund will own a portfolio of high-tech game-changers and disruptive business models using advanced science and engineering to tackle the toughest and potentially most lucrative tech challenges.
Max Accredited Investor Limit: 249
Deadline to Join: December 31, 2023
In this Deep Dive, Grant discusses:
- Home
The significance of analyzing consumer content and the trends to anticipate
- Home
How a framework to organize content can help identify and mitigate potential risks
- Home
The future that generative AI can enable
- Home
The importance of prioritizing infrastructure investment before the widespread use
Want to learn more?
View all our available funds and secure data rooms, or schedule an intro call.
New to AV?
Sign up and access exclusive venture content.
Contact [email protected] for additional information. To see additional risk factors and investment considerations, visit av-funds.com/disclosures.
*Michael Chui, Et al., “The Economic Potential of Generative AI: The Next Productivity Frontier,” McKinsey Digital, June 14, 2023
Frequently Asked Questions
FAQ
Speaker 1:
Hey there, I’m Grant Demeter. I’m an investor and a principal with Alumni Ventures, and today I want to talk about generative AI and the future of consumer content. First, a brief definition: generative AI today, I mean, we think of that as things like DALL·E or ChatGPT or other large language models or image generation models or other new models that are coming out. Together, they really generate—right now—content: words, images, video, environments, things like that. And when something is generating content and it’s so democratized, there’s definitely going to be some kind of downstream impact on the tastes, preferences, and behaviors of the everyday consumer. This is a foundational change. And so one of the questions that I want to try to answer—or try to start to answer—in this piece is: how is generative AI going to influence the future of the consumer content industries?And as part of that, we’re going to have to define what is content? What did the consumer content industries look like up to today? And what trends might we be able to pull through to the future to try to understand what might tomorrow look like? So that’s how I’m going to try to approach today’s conversation. A quick orientation: I’m going to spend quite a bit of time trying to define the space and lay out an investment thesis informed by theory and by a little bit of history. Then I’m going to talk about some hypotheticals around what needs to be true in order for these spaces to play out as I think they might. What would indicate success or failure in this future? A couple of companies which I think might be driving towards this future, and then some resources for us to continue to learn and follow as we continue to evolve our understanding here.
I will say that this piece today is an abridged version of a blog series that I write called Markets by Grant. I did a three-part series on this. If anything kind of falls between the cracks here in this more brief overview, feel free to give me a read. I would appreciate it. Alright, let’s dive in. So I’ll start with the immortal words of George Costanza: in order to manage risk, first you have to understand risk. How do you spot risk? How do you avoid risk? And what makes it so risky? And the same is true of any industry, right? In order to understand any industry, first you have to define it. What are the characteristics of that industry? What makes it good, what makes it bad? In this case, I think content is one of these really amorphous, nebulous terms that has this industry or industries that kind of float alongside it.
But no one can really say exactly what it is. The word “content” is kind of like “matter” or “media,” and people use these words super interchangeably. Is a book content? Is a movie content? Is a podcast content? Is computer code content? Where does content begin? Where does it end? And how do we organize and classify it based on its characteristics? And I started off as a content investor and a consumer investor trying to find different frameworks online via desk research to try to help me understand and organize my thinking around it. And was disappointed to find these kind of screenshots that you see here, which are not MECE—not mutually exclusive or collectively exhaustive. They’re just kind of word-cloudy combinations of different words, which together might give you kind of a loose idea of what content is and how we should think about content and its characteristics.
But really, I’m going to go on to show that content is kind of like the underlying product of most consumer industries today. And I talk about that more in my blog as well. And I think that that product—if we’re really supposed to predict how generative AI, which is fundamentally a content innovation technology, is going to innovate the future of content—well, then we need to understand and have a point of view on what exactly it is and how it’s organized. So I’m a big framework guy, and some of this is going to be pretty theoretical, but I wanted to start off by creating my own framework for how we think about defining content. How has the content industry evolved over time and what does that mean for the future? So bear with me as we’re going to start pretty high level with some definitions.
My most basic definition of content is: information that humans consume and decode using our senses. So our senses—touch, taste, smell, sight, sound, etc. I mean, an example of touch-based content is Braille. I could have included touch here, but for the purposes of keeping things relatively simple, I distilled it to the two main ways that we consume content, which is visually with our eyes and auditory with our ears. So content that we hear and turn into information that we discern; things that we see, synthesize, and come to understand—that is how I’m defining content. And on the left-hand side, we’re going to see some characteristics which define the persona of a person—in this case, a consumer—experiencing that content, and some characteristics of what that content itself is like. So the first part of this framework is the two types of content: visual and auditory content.
And then you see the word “static” and “passive.” So my first bucket is static content in visual and auditory. This is content that you look at—that it is what it is, right? You look at a piece of paper with words on it—that’s static content. You’re not going to look away and look back and the words are going to change—not at least with a traditional book or article or other piece of text-based content. The same is true for images. And these images could be photographs, they could be drawings, they could be anything. All this, if you boil it down, is just basically synthesis of visual symbols which we use to encode and decode information. Sometimes this can be purely informational, like reading an article. Sometimes this can be aesthetic and artistic, like looking at a painting. But the mechanism with which we consume the content and our persona in consuming the content is the same, which is: we are passive consumers of static content.
Now I’m going to introduce a new type of content, which is temporal content. This is content that’s not always the same. This is content that plays out over a specific period of time. And this is most forms of modern content that we’re likely all aware of, like TV shows, movies, podcasts, music, etc. And as you can see, over time in most modern forms of content, there’s been a confluence or convergence of visual and audio. And most content that we consume today—at least most modern forms of content that we consume today—are both audio and visual. So even podcasts and Spotify—you see that they’re putting visuals to them. Or music in Spotify—the concept of the music video, television, etc. This is modern content today. But it’s not the most modern form of content. We’re still passive consumers. We sit down, we consume this content, the content plays out, it changes over time, it’s done. We can revisit it—the content will be the same.
I’ll introduce a new category of content based on my framework, which is what I call experiential content. Now this is things like gaming or social or AR, VR. And I understand that gaming and social may not feel like they make sense in the same bucket because they’re very different in terms of user immersion and the reasons why you might engage in it. But I really do think that if you distill them down, they both belong here. Experiential content is a content kind of mini-world that you’re able to manipulate as you experience it. So you’re not able to define it, but in a rules-based way, you can explore this content, you can choose, and those choices will lead to different content experiences. Now the only thing is that in this environmental paradigm, the number of content experiences that you can have—while it’s a whole lot more than with, let’s say, a movie or a TV show because it’s choose-your-own-adventure—it’s still relatively finite. In a video game, there are only a certain number of worlds that I can explore.
There are only a certain number of conversations that I can have or weapons that I can use. In social media, there are only a certain amount of other people on social media. There’s only a certain amount of combinations of words that I can do. And there’s only a certain amount of functions that exist within the social media platform. This is what I call postmodern content. This is today’s content paradigm, and you’ll see that we’re no longer passive—we’re active in our content experience. We’ve developed the muscle memory of wanting to co-define the content which we’re experiencing, rather than sitting and just letting it wash over us. That’s the muscle motion of the modern consumer.
But then this new thing came around called generative AI, and what appears to be a new cycle has started. But it’s that same active muscle memory which is informing our content experiences, meaning I am actively defining the types of text, image, video, audio that I want to consume—which means that there’s this tighter interplay of content creation and content consumption, which is truly a new paradigm. It’s an active experience. But right now, we’re seeing a cycle restarting in that we’re starting in generative AI with kind of old-school forms of static content. You can create images, you can create text, you can create short-form videos, you can create environments. But we’re not quite to the point where we’re actively defining experiences or temporal worlds that we’re existing in and exploring at the same time as creating them. And I’ll get to that later, but I’ll pause here and say that the point of this is effectively to say that the future in some way is this idea of co-creating, co-consuming content at runtime, which effectively means while this content is being consumed, we are actively defining the direction of this content—outside of a rules-based paradigm or in less of a rules-based paradigm than experiential content.
And we’ll get to this later, but some terms that may be familiar here are “metaverse,” which is an example of the beginnings of this paradigm where you create your own digital world, digital identities—the boundaries of which, and the rules of which, are defined by you. And although this seems very high-level and kind of pie in the sky, I hope that this will start to make more sense as I go on. So let me close this slide by saying: this is my super theoretical, high-level view of content and how it’s trended. But what I wanted to do next is see if I could validate that theoretical view with actually what has happened in history. And as part of those historical technological innovations, see what the future might look like. See if I can continue to pull some trends through to the future. So high-level: content has evolved from static to temporal to experiential.
Now generative—not mutually exclusive. We still consume all these forms of content, but these have been the innovations. And as part of these innovations, the consumer has moved from more passive to more of an active role in consuming and creating this content. Alright, so now we’re going to go back into ancient history and see if a historical view will validate or invalidate my point of view. So as you can see, this has some similarity to the previous slide in that it’s a funnel shape and there’s some color similarities here. And I won’t ask that you have to digest all this information—it’s highlighted in greater detail in my blog—but effectively it’s highlighting the major inventions in the realm of content. So: content technology inventions and new forms of content which they enabled. And I want to highlight a couple of trends on this slide.
Speaker 1:
You’ll see that static content and temporal content are here too, and we’re going to talk about that. You’ll see that I’ve highlighted on the right-hand side some trends. I’m going to close by talking about that, and you’ll see that I’ve also kept a differentiation between auditory content with the green lines and visual content with the blue lines. But I wanted to highlight something else, which is that over time you’ll see that there’s a lot of these semi-parallel lines, which indicates that back in the day—and we kind of take this for granted that this isn’t the case anymore—content capture technology and content consumption technologies were disparate. For example, when the printing press came out, we didn’t use the printing press to read the Bible. We used the printing press to create the Bible. And then on a totally separate medium, we read the Bible.When the camera came out, we weren’t able to see the photo that we took. We had to take it through this laborious darkroom process, which took days and different forms of media. And then we ended up looking at a piece of paper days later, which is the content that was actually captured using this camera technology. So back in the day, these were very disparate. And as a result of that, it wasn’t super democratized to create content. It made sense why we were passive consumers of static content, because rarely were we able to iterate on content so quickly that this very same content which we created we could choose to enjoy, and so on and so forth. So it was kind of a professional trade to create the content and to sell that content to consumers. Over time, there are three trends which changed this: technology consolidation—
So that is, effectively, from camera to photograph. We continue to get closer and closer. We have a film camera, which means that we can get to photographs quicker—in an hour. Then we have a digital camera, where we can look at the photograph immediately on the camera. Another one is the democratization of content creation. As these innovations in cameras, as an example, continued to move forward, we found that they became cheaper and cheaper and more accessible for everyone, which means that in general, more content was being created by more people. People’s tastes were evolving faster than ever, and people were iterating faster on content and innovating faster on content. And the result of that is the third trend, which is popularization at large. Right? Like back in the day with the printing press and early forms of the camera and photograph, content consumption was looking at a picture on the wall or engaging with your Bible at home.
Now, content experiences are all around us. We spend most of our time, if you think about it, in some form or another consuming content. And that’s because content has popularized aggressively over the past couple hundred years and democratized aggressively and consolidated aggressively, and now it’s consolidated effectively into one device, which is the personal computer. And now I say—call that 1980. In 1980, with a personal computer, you could both create and consume all of these different forms of content—static and temporal—all in one place and one interface. And that was a true paradigm shift in consumer technology, which enabled the shift to experiential content. The PC and its kind of rules-based interface and setup enabled us to transition from static to temporal to a new paradigm of experiential, rules-based content experiences. And that was also the transition from hardware, obviously, to a software paradigm of content. Instead of interacting with these physical photographs or physical books—which we still interact with—more and more, we’ve started to interact with content using digital methods: using screens and using speakers…
Speaker 2:
And—Speaker 1:
Other forms of digital technology enabled by software. And as you can see, I kind of got lazy here and I didn’t want to draw out the whole history of gaming and the whole history of social media, but these are two industries and forms of content which were made possible by the personal computer. And they continued to evolve in the same directions, which is accelerated popularization, accelerated democratization. Social media—especially in this idea of UGC or user-generated content—brought this about. But then one more thing that we saw is fragmentation of applications. We started to see many, many, many more games pop up, many more types of social media pop up. And that’s because this PC innovation allowed this massive wave of invention that was so democratized that the market was flooded with different ideas. And that brought us all the way up to yesterday, I’d say, where the metaverse and Web3 forms of democratized, user-generated, and rules-based content—which are on the bleeding edge of rules-based and starting to approach kind of infinite creativity—Were starting to become a really hot investment theme and area. And then all of a sudden that kind of went on hold for a little bit, at least from the lens of the traditional VC, and we got into generative AI. And so here is a similar funnel lens for generative AI. You can see that it kind of mirrors the trends of the above static and temporal content, which is: we start with text, then images start to be able to be created; we start with voice, then music starts to be able to be created. And we’ve gotten down here where we’ve seen many more forms of content being created and in fewer places, models starting to get democratized, and innovation—and the pace of innovation—happening much quicker. And of course, this isn’t to scale. I mean, this has all only happened in the course of five or six years. So this is, as you can see, really accelerated. And so just wanted to kind of compare this with what we were looking at before, which is: I think overall, this idea is kind of validated—that the content consumer has transitioned from passive to active. The forms of content that we experience are transitioning from static, time-based, experiential to generative. And with generative, they’re also cycling through those categories of static, temporal, and next would be experiential.
And so what would that experiential paradigm look like? Well, if we take the trends of yesteryear and try to extend them to today, we see that content technology—content creation technology using large language models and image models—is starting to be a little bit centralized and evolving at a faster pace, and it’s becoming cheaper and it’s becoming more consumerized. I think that the next paradigm is this idea of the personal generator—or I’ll call it the PG, or coin the PG. And the concept of the PG is it can do everything. You can have all the types of content experiences that you have on your normal personal computer, but the difference is you are co-creating and co-experiencing all that content at the same time. Instead of watching a TV show, which might be designed for your personal demographic, you can create your own TV show. Instead of listening to a podcast, you can create your own podcast. You can change the podcast as it’s going on.
You can write your own story, and you can constantly evolve that algorithm faster than it can be evolved for you by other traditional forms of content creation. But outside of those forms of temporal content, you can also start to create your own games, your own social experiences, your own worlds, and so on and so forth. And that is what I think the real future is. It’s a return to this concept of metaverse—but very much user-defined, user-generated, and enabled by generative AI. And there’s a lot of small words here. I would take a look at the blog if you’re interested to learn more and try to digest this a bit more. And I will say also as my last disclaimer for this slide, that a lot of this stuff is already out of date and it’s certainly not exhaustive. Alright, so that was a lot.
So I just wanted to highlight what I think might be coming next in this world of consumer content slash generative technologies. And that’s a continuation of some of the trends that we saw: technology consolidation, democratization of content creation, and popularization. We’re still seeing those trends play out. And then some trends which also need to play out—and I believe are going to play out—are: social gamification, user immersion, expansiveness of content, and personalization of content. And let’s dive into those. So in order for those to play out—in order for us to reach this concept of the personal generator—we’re going to need to see technology consolidation continue to happen at an aggressive pace. So that means more content generation models being placed in smaller and smaller individual places and starting to be able to reach you on various forms of technology on an everyday basis.
So from the PC to the mobile phone to the VR headset to your watch, etc. We’ve already seen that democratization and popularization of generative platforms is blisteringly fast. I’m going to butcher this metric, but we basically saw that ChatGPT got, I think, to a million active users within a week and 30 million within a month. And I mean, that is pretty unprecedented for technology today, but just shows the continued democratization and popularization of content and generative platforms. Social gamification—I think we all know that’s here to stay. Social media continues to innovate. Social theory around social capital exchanges continues to prove out in terms of the value of creating a social consumer experience is amplified when there’s gamification. User immersion is the future. I think this is one that we all understand, which is: we want to see more immersive, realistic content that draws us in more and more. And there are two AI enablement layers which can make all this happen a little bit faster. One is expansiveness and extensibility. So AI being able to create things faster than anything before means that the breadth of this content is much larger and can be created in a much shorter period of time. And this also means that content can be personalized much quicker to individuals.
So taking a couple steps back here—what does this mean for how we invest? Well, the first thing I’ll say is: we’re definitely not there yet. If we were there yet, I wouldn’t be opining at such a high level. We’re not there yet. And so since we’re not there yet, but we want to invest theoretically in this future, there’s a couple of things that I might say we should think about. And one is: investing in the infrastructure before the application. And I think this is especially important for a consumer market, because in the consumer markets, if you come out a little early with something which may be really resonant 10 years later, you’re not rewarded for it. But who is rewarded is the enabling technology infrastructure who allowed all this to happen from the beginning—because they can ride the wave going forward. But I will caution to say: the infrastructure isn’t fully what you think it is.
The infrastructure in this case is not just the large language models or not just the chips. The infrastructure is the social infrastructure as well. It is the social platforms that have incumbency of users where generative AI can be introduced or new forms of content can be introduced. And as part of that, I’ll say that consumer science is just as important as data science for the consumer industry. For consumer industries, we find that the burden of retention is much higher, and that burden of retention really needs to be proved out in a simple, elegant, undeniable customer value proposition. And that can be just as hard as creating a significantly outperforming large language model. And so my summary is that timing is everything and patience is key for investing in the consumer content space—especially in this fast-moving generative AI paradigm. A couple hit-or-miss things to take a look at.
Speaker 1:
One is Apple Vision Pro. I think a lot of us saw the announcement and thought that this could be a paradigm shift, especially its App Store, because if you’re selling apps on top of Apple Vision Pro, you’re selling apps for a new paradigm of content consumption. If that is popularized, that will accelerate our move into this new future. The metaverse returning—or metaverse re-popularizing—would also be an accelerant of this shift. Existing consumer social platforms getting generative—we’re seeing this with Snap, certain Meta platforms, and others.Regulation is an important topic here. If regulation preserves democratization of content creation and limits the regulation of the images that you create, or the text that you create, or the videos that you create, this will enable the whole system to move and cycle faster. And then we also have to believe that the supporting infrastructure continues to make things faster and cheaper and more mature.
And then quickly, these things could be a slowdown on the right side. So if regulation puts up high guardrails, if Apple Vision Pro ends up flopping because it was too early, if the predominant culture considers generative content and content creation to be inauthentic. And then if we start to introduce things like paywalls and tech infrastructure bottlenecks, and we see some limitations to innovation there—some things to keep in mind to determine whether investing would make sense now or later.
Then I wanted to highlight just a couple interesting startup ideas as I’ll move towards closing things out. Not all necessarily investable, and certainly all early in their own paradigm, but I’ll highlight a couple examples of each. So, as I mentioned with this concept of this personal generator, we can generate a lot of things. One of the things that we can generate is worlds or metaverses. And generating worlds makes a lot of sense intuitively in platform gaming environments.
So there’s a lot of these professional tools which are popping up, which are using generative AI to help professional game developers build worlds for their games—seemingly infinite worlds for their games just based on verbal commands. One is called Promethean AI, one is called Atlas Design. There are a few more. And then there’s some very rudimentary, fun ones to look up—not necessarily investable products—which are doing this for consumers, where you can speak a world into existence and you can say, “Build me a world with a beach and a boat,” etc., and then it’ll start to build you this world in real time as you co-experience it. One’s called Moatboat, which is now sunsetted. You can still find it in some places, I think. And then another is Builder Bot, which is a Meta AI product, which was demoed a couple months ago.
The next category of things to generate with your personal generator is minds. And minds I think of as companions and characters. So this is familiar for those who have seen kind of modern films today where there are kind of robotic or AI companions to the main characters. And this is already starting to be able to exist with startups like Character.ai, Replika, Chai, and Anima. Basically, you can build personalities who you can share life with, who will learn you and empathize with you and come to understand you, and who you can converse with on a daily basis. Then in parallel, you can also create more independently operating characters. And gaming is a key use case for this, where you can create non-player characters who have their own personalities and own motivations and motives.
This is kind of like a Westworld paradigm where you create this character and it will run and live its life independently of you, and you can still interact with it. This is obviously much more nascent, but a couple of startups in the space are Inworld, Charisma, and Kinetix—largely building for gaming use cases.
Then there are platforms. I mean, these aren’t startups, but these platforms are going to be, I think, where you’re going to be able to experience these new forms of generative consumer content. And this is metaverse platforms, social platforms, and then physical platforms like the Apple Vision Pro. So I mean, a fun kind of meme example is buying the Vision Pro to see my AI girlfriend, right? The Vision Pro is the platform with which you have the consumerized generative AI experience with a companion—like in this slide.
And then the last concept is very horizontal innovations of generating anything you can imagine. And I mean, there’s a bunch of companies playing in this space, and it’s tough to even call them startups because they’re so well-capitalized and mature and they’re very familiar to a lot of people. One is Runway, which has a large variety of models building things on the frontier such as video. And the other is, of course, OpenAI.
So I will now wrap it up. Thanks for sticking with me for a while here. A couple different sources if you want to learn more. I breezed through it—at least I felt like I did—but my blog is a little bit more detailed and measured in defining and explaining what I tried to voice over effectively today. So feel free to give it a read.
Andreessen Horowitz has done some great writing on generative AI, which partially inspired this piece. And then NFX is always—I always see them as fairly early to doing writing on different and emerging topics in generative AI, so they’re a great source to follow, another venture firm.
Rex Woodbury is a consumer tech writer and a VC investor, and he has a weekly newsletter called Digital Native, which I think is really well done. And then I think a foundational thought piece to understanding how consumer social economics work and how to design an effective consumer app—back when I said that understanding the consumer is just as important as building the right technology—is this piece called Status as a Service by Eugene Wei. And it’s a long piece, but I think it’s absolutely fundamental for understanding the space and understanding how to invest in consumer businesses—especially consumer businesses which are on the vanguard of new technologies such as will emerge here with generative AI and content.
So quick summary: generative AI—I do believe it’s the beginning of a paradigm shift for how people will create and consume content. The personal generator is going to enable us to simultaneously create and consume content, and that content is going to be social still. It’ll be immersive, expansive, and it’ll be personalized to each one of us.
It’s still very early and we don’t have a whole lot of signals to go off of, but we have historical trends and I’m hopeful that some of these historical trends will validate the future vision—the future vision thesis that I’ve laid out. And of course, the consumer is still the consumer, and we can lean on our understanding of the consumer as we think about how to build for them with new technology.
And we’re going to use these lenses, and we’re going to evaluate today’s influx of applications and infrastructure and new forms of distribution to make the right investments. It’s a really exciting time to be a VC investor and thinking about all this stuff and meeting with a bunch of people who have thought about it much more than myself. And I just feel like the learning cycles are so much faster today in consumer because of generative AI.
And it’s been a pleasure to be along for that ride. And thanks for sticking with me along for this ride. Thank you.