How Many People Actually Use AI
Nate Elliott · March 5, 2026
ai in marketingseogeo
**Welcome & Guest Introduction**
(0:04) Hey everyone, I'm David Burkwitz. Welcome to another edition of AI Insiders with AI Marketers Guild from Architecture Media. I've got a guest; this one's been decades in the making. Nate Elliot is someone whose research, analysis, and insights I've been following for many years. We were, I guess, even quasi-competitors, or at least our firms
(0:32) were sometimes got along better than others and were navigating how to play well together. But he's someone I've been able to admire in that healthy competitive sense, from everything he's been doing, one of the ad and marketing industry's top analysts. Then once he joined my old home, eMarketer,
(0:58) where I spent a lot of my formative years in the industry really early on, and seeing what Nate is doing covering AI in certain areas as part of his purview. Nate was able to come to my office, which is the coffee shop in Koreatown, and we got to catch up on things. I'm just so excited to geek out here today and to get to have
(1:26) conversations with one of those people who's one of your gurus, who you actually get to meet and learn from. I'm like, Nate, as soon as you've got something ready to share, come on AMG. You're going to get some way better questions than I'll ask from this crew. If you haven't been here before, for everyone else, these are interactive conversations. Nate, I could go on for the whole hour welcoming you, but that wouldn't be fair to our crew here. Welcome.
(1:47) Thanks very much. Earl, I think you were clapping, David, saying you were going to stop. Is that what you're...
(2:08) Oh, yeah. That was me. That was me clapping. I will neither confirm nor deny. Excellent. Well, thank you, David, for having me. I'm excited to be here. Thanks to everyone else for joining today, both live and anyone who watches the recording. Hello, future people.
**Nate Elliott's Background & Focus**
(2:19) My name is Nate Elliot. I am a principal analyst at eMarketer. I cover how AI is changing marketing, commerce,
(2:27) and the customer journey. As David said, I've been an analyst in the industry for quite some time. I actually worked at DoubleClick back in the 90s when we were helping to invent the way advertising would work, for better and for worse. I joined Jupiter Research as an analyst in 2003, and since then I've been an analyst ever since, at Jupiter, then Forrester. I ran my own shop for
(2:48) about a decade, and I was thrilled to join eMarketer last summer and get to work as their lead analyst on AI. What I want to share with you guys, and forgive me, we're a Google shop, not a Zoom shop, so I'm going to ask you, David, to tell me if I'm getting this right as I share my slides. I still don't know how to keep an eye on the Zoom chat while
(3:12) I'm presenting. So, I'll surface if someone's shouting things out. What I want to talk about is the building blocks of AI adoption. I've been looking at this space even before I came to eMarketer to lead the coverage of AI.
**The Quest for AI Adoption Truth**
(3:20) Of course, I was paying attention. I was running surveys and research on this for companies, including Walmart and
(3:34) Burger King, and a bunch of others. One of my first tasks when I got to eMarketer was to create our single version of the truth around what's happening in AI adoption. All the stuff we want to talk about, the analysis we want to provide to platforms, to vendors, to brands, be they marketers or sellers, all of that is analytical. I would dare
(3:56) say there is no truth. There are simply informed opinions, and that's where the job gets really interesting. As a baseline, one of the things I wanted to understand was how many people are actually using these tools? Because it's a thing we talk about all the time. It's a concept, a number that I honestly don't believe the industry has put really good
(4:20) analysis or factual data around.
**The Discrepancy in AI Usage Data**
(4:20) I see numbers that are absolutely all over the place. One of the things that I like to do when I'm in a room with people is ask everyone to shout out what percentage of the population uses AI in the US. I did this at a couple events last week. I got numbers as low as below 20% and as high as 85 or 90%. The joke, of
(4:46) course, is that whatever number you say, as long as you're between, say, 10% and 99%, you're correct because I can find data from a credible source that will back you up, that will say that yes, that is the number of people who are using it. To answer the question today, I want to use a piece of research that was published a little over a year
**Gallup Study: AI Usage Awareness vs. Reality**
(5:00) ago. It was conducted in December of 2024, and I'm fully aware of it. Showing you AI usage data from 2024 is like showing you space travel data from 1944. It doesn't make a lot of sense, but hopefully it will in just a moment. This is a study that Gallup published last January 2025, that ran in December of 2024, and they asked about 4,000 US online users, "Do you recall
(5:39) using an AI-enabled product in the past 7 days?" They weren't just asking about ChatGPT and Gemini and things like that. They were saying, "Is there anything else you've used that is AI-powered or AI-enabled?" And 36% of US online users said that they could recall having done that in the past week. Then Gallup came back and said, "While you're here, let's just ask if you used
(6:01) any of the following tools."
**99% Use AI (Known or Unknown)**
(6:01) It turned out that 99% of these same exact respondents had in fact used one of these AI-enabled technologies in the previous seven days. Everyone uses AI at this point,
(6:26) whether they know it or not. That's been true, of course, for at least a year or 15 months, probably longer than that. I'm sure that I'm preaching to the choir a little bit. But one of the reasons I want to show this, even to an educated and interested group like the AI Insiders, is to reinforce this gap between people choosing to use AI, or being aware of using AI, and actually
(6:50) using or interacting with these technologies. That gap makes a huge difference to the brands, the marketers, and the sellers who are looking at using AI as a marketing channel. Helping those brands, marketers, and sellers figure out how they can best leverage AI as a marketing channel is a big part of my remit. It's a big part of the work that I'm doing.
(7:12) The thing is, not all AI adoption is created equal. While lots of people use AI, 99% of people in that study from a year and three months ago had actually interacted with an AI tool in the previous month.
**Adoption vs. Adaptation in AI Usage**
(7:12) Not all of them had chosen to do so. Even amongst those who choose to use AI, the adoption is really high, but the adaptation, as my colleague Jacob Bourne
(7:37) would call it, is really low. What we talk about with adoption versus adaptation is adoption are people who use the tools. They go and they choose to take advantage of the tools. Adaptation, we're talking more about changing behavior patterns: taking things you've done for years online and moving those behaviors from one tool or platform or location to another tool or
(8:00) platform or location. One of the big behavior patterns that we see, of course, is people going to traditional search engines, Google. The number one reason people go online is to find information. That's been true for a couple of decades now. Google has, depending on the market you're looking at, anywhere from 70 to 95% market share in traditional search.
(8:21) When we look at traditional search engines, we know that 95% of people who go online every month are using a traditional search engine at least once a month. And 86% of online users, almost all the people who go to a search engine monthly, are going there on a frequent basis, 10 or more times per month, at least about once every third day.
**AI Tool Adoption: A Sampling Phase**
(8:42) When we look at top AI tools, the numbers look a little bit different. It's much earlier in the adoption curve. So the total adoption, the total number of monthly users, is not going to be as high. 38% as of August of last year is still a remarkable number for a set of tools that at that point were less than three years old as a category. What's really interesting to me is that
(9:05) only about half of that 38% were choosing to go to these tools on a very regular basis, 10 or more times per month. A lot of people are playing with AI. They're sampling AI. They're using it unintentionally, or they're using it intentionally every now and then. While we all live in this bubble where everyone uses AI all the time, if you look at the general online
(9:29) population, the reality is a lot of people use AI a little bit, and a handful of people are using AI a lot. That distinction, the gap between those polls, is really interesting and important, I think, to the marketers and sellers who are looking at using AI as a channel.
**Is AI Mobile-First?**
(9:48) Nate, quick question here.
(9:51) I mean, AI just by its nature also strikes me as something that is so mobile-first. Do some reports like this give short shrift to that concept, that you want AI to be with you everywhere?
(9:51) I'm sorry to cut you off. I wouldn't say it's mobile-first. It's snackable interactions,
(10:15) and that's often on mobile. I do a lot of my stuff on desktop.
(10:15) Oh, I do too, but I feel for most who are creating a recipe or telling a kid's story...
(10:15) That's a fair point. You're right.
(10:15) No, but both can be true.
(10:34) Mobile's a huge part of the story, if for no other reason than mobile's a huge
**Challenges in AI Usage Data Collection**
(10:34) part of the story of how people use all digital technology at this point. The study that I'm showing you happens to be based on desktop data analysis and clickstream data. Again, it gets to my point that it's really hard to find a good single source of truth around how many people are using these tools, which tools they're using, how frequently they're using them, and for exactly what
(10:53) purposes. There are imperfections in methodology no matter what study you look at. There is no research methodology that's going to perfectly give us this data. This is an imperfect study alongside a lot of other imperfect studies. I will soon be publishing my own imperfect studies, but hopefully in a way that add
(11:15) to the clickstream observational data, like what you see on the screen right now, to add some nuance to the conversation and to fill in some really important missing gaps that we see not just in studies like this, because Semrush and Dallas did a great job on this research, but it's limited by the data with which they were working. So I want to start to fill in those
(11:38) gaps with a consumer survey methodology, and we'll talk about that in a moment.
**OpenAI's Data Dump & Challenges for Marketers**
(11:38) People use AI in lots of different ways. This is great data. Full credit to OpenAI. The study they published called "How People Use ChatGPT" in September of last year is one of the greatest corporate data dumps I've ever
(12:02) seen in a quarter century of being an industry analyst. They update a lot of it, not all of it, but a lot of it on an ongoing basis for us, which is incredibly generous, and is one of the handful of things they do that still reference their founding as OpenAI, a company that would share its knowledge and information with the world. I show this
(12:23) slide because this is really hard to process. It's, I think, seven different primary categories, two dozen different subcategories, and there's this enormous range of different ways in which people are using artificial intelligence. Brands and retailers look at this, and they come to companies like eMarketer, and they say, "What am
(12:47) I supposed to do with this? Where am I actually supposed to start if I'm hoping to reach people who are using these channels?" I'll get it out of the way upfront. There's an entire section of this industry that is not using AI as a channel, but is using AI to make your work better as a marketer, a seller, a brand, a retailer. That is research that
(13:07) we are working on and will be producing. But a lot of what I'm talking about today is how people are using AI, and therefore, how brands and sellers can use it as a channel to reach those people. When we talk to brands and sellers, we think, rather than look at this enormous amount of data that's out there, rather than look at slides like that colorful spaghetti I
(13:26) just showed you on the previous chart, ask these three questions, and ask them in this order. First and foremost, how many of our customers use AI?
**Customer-Specific AI Usage**
(13:26) That's a more difficult question than a lot of us probably understand because in this little bubble that we all work in as AI Insiders, and people who think a lot about this, we assume that the vast
(13:49) majority of the world uses AI. You hear things like AI is replacing Google. I've seen people stand in front of slides that say 80% of the population goes and uses LLMs every month, and I've never seen really credible data that says that's true. But I have seen data that says that's true. Whatever the overall number is, I think the first thing that brands
(14:13) and sellers have to ask themselves is not how much of the general population uses AI, but how many of your customers use AI?
**Generational AI Usage: Most Adoptive**
(14:13) I want to play a game with you guys. I'll ask you to pop it in the chat or shout it out, but which of these generations do you think uses AI the most in the US right now? Any guesses? David, what's your guess?
(14:35) I want to say Gen X, but I feel millennials are still going to be a little more tech-forward. So, I'll say millennials.
(14:35) We'll say millennials. I see people go in the chat. I'm seeing a bunch of different guesses in there. Gen X, Millennials, a lot of people are focusing on Gen X and millennials. Those are some solid
(14:58) guesses. The actual answer is category D, Gen Z.
**Generational AI Usage: Least Adoptive**
(15:00) Let's flip that coin around though. That's the population, the generation that uses AI the most as a percentage of online users. Which generation uses AI the least? David, let's hear your guess. I know you have a guess on this.
(15:00) Well, Gen Alpha.
(15:00) You think Gen Alpha?
(15:20) Yeah.
(15:20) All right, let's see what else is going on in here. An even mix of boomers and Gen Alpha. A couple people said maybe it's Generation X. The generation that uses AI the least, you guys have gotten this right. Congratulations, more than any other audience I've asked this question. It's
(15:42) actually Gen Alpha. The reason is this: a lot of Gen Alpha is online. We run these numbers in our forecasts as a percentage of online users. That's what I'm looking at here. Something like three-quarters of Gen Alpha is online. That's terrifying to me because Gen Alpha right now is between the ages of two and 13 years old. For 75% of them to
(16:06) be online is scary. Our forecast team assures me that if a parent starts YouTube on their tablet and hands it to the toddler in the stroller, that is in fact an online individual. That's how we get to 75%. The reason that those online users in Gen Alpha aren't using AI is that most of them don't know how to read and write.
**Gen Alpha's AI Adoption & Future Projections**
(16:33) I do wonder, and this is my own bias here, having a 12-year-old, that my 12-year-old is so skeptical herself. I feel there's, and we often see this, generations back-to-back often one rebels against the other. So if my daughter sees all these college kids and 20-somethings obsessed with AI, it's like, screw them, right?
(16:56) Yeah. There are definitely a lot of motivations that we didn't get into because, to answer Selena's question, this isn't a study, it's a forecast. This is a model that was put together by our forecasting team using thousands of different data points from hundreds of different sources, then modeling it forward based on how we see the market
(17:16) changing over time. We do think that once these two to 13-year-olds—I'm assuming, David, that your 12-year-old is very good at reading and writing, but the 2-year-olds certainly don't know how to read and write. The average age at which someone in the US learns to read and write basic words is 6 years old, I believe. That means over the next few years, as Gen Alpha gets more
(17:38) literate, they will start to use AI more, and we're forecasting that they'll overtake boomers in 2027 by the end of next year.
**Gen Z Leads AI Adoption, Challenging Assumptions**
(17:38) Anyone who guessed millennials as the biggest adopters of AI, you would have been right about a year and a half ago. But according to our model, Gen Z overtook millennials towards the end of 2025. The reason I love your
(18:04) comment, David, your 12-year-old called Crusty, that sounds right, not because of you, but because of Gen Alphas. The reason I show this and the reason I play this game, which hopefully you guys found fun, maybe enlightening, is to show you that we are some of the most interested people about AI adoption and how AI is changing our world and our
(18:29) industry. A lot of us got this wrong. I would have gotten this wrong if you'd asked me this question before I sat down with the eMarketer forecast on this topic. We make a lot of assumptions, and a lot of those assumptions are based on what the people directly around us are doing. The people around us are not the general population, and they certainly don't, on average,
(18:50) reflect the entire spread of individual groups, be they generational, gender, race, income, education, whatever it is. The people immediately around us don't often give us the best lens to see what's really happening in the world. That's why we think the first question that is really important for marketers and sellers to get great data on is what percentage of your audience
(19:12) specifically is using AI, because that will start to tell you how important AI is in your marketing and your sales strategy.
**AI's Role: Primary or Secondary Channel?**
(19:12) I don't think there's a seller or a brand, no matter their audience, that shouldn't be working with AI. But I do know that there are some sellers or brands for whom AI is perhaps the single most important channel for getting messages out and marketing to and
(19:37) selling to the people that they're trying to reach. And I know that there are a lot of brands, probably many more, for whom AI should remain at this point a secondary channel that they're leveraging to a limited extent, while keeping their focus on the larger, more important channels today, and also testing and learning and preparing themselves for when that tipping point
(19:59) eventually comes. It's a really important distinction because I've talked to marketers, I've talked to CMOs who think we need to drop everything and only focus on AI right now. There are a handful of companies and product categories in which that is true, but that is not what is happening for the vast majority of brands, fans, marketers, and sellers
(20:22) right now.
**Understanding Customer Motivation for AI Use**
(20:22) I think understanding exactly how much your audience is choosing to adopt these tools is a really important starting place that a lot of the people I talk to aren't actually starting their investigations. Once we answer that question, I think the next question we have to answer is, why do our customers use AI? When I talk about why, I want to
(20:43) be clear. I'm not talking about what features or tools they're going to these platforms to use. Not, "Hey, the why is they want to play with Sora 2," or, "the why is they want to play with a character AI." Those are important things. We're going to get to that in the third question, but I want to know why people are actually choosing these tools in the first place. Why are
(21:05) they going to these platforms? What is the human motivation that drives them there?
**SEO is Not Dead & Mixed Data Challenges**
(21:05) To Adam's question in the chat, no, SEO is nowhere near dead. We can talk about that in a little while. But when we talk about why, most of the data I see, and again, I'm having a really hard time finding single sources of truth on this, most of the AI usage data I see, whether it's
(21:26) survey-based, whether it is behavioral and observational, at the very best mixes motivations and features. That OpenAI slide, that rainbow spaghetti I showed you a few minutes ago, talks about why. And Daniel Green, yes, there are some Anthropic and OpenAI reports on this. But just like the AP survey on the screen right now, just like the OpenAI data I
(21:50) showed you a few minutes ago, what we're really seeing is a mix of motivations and also the technologies, the features, the actual tools that they're using. I don't think the tools are as important a question as the features. There are a couple of reasons for that, but let me skip ahead and get to that.
(22:13) The third thing is, which tools do our customers use?
**AI Tools & Features: Rapidly Changing Landscape**
(22:13) First, how many of our customers are using AI? Second, what is their motivation for going and using these tools? And the third question is, which tools do they use? The reason I don't think the tools and the features are as important is because they're changing every day. There are days in
(22:32) which multiple frontier platform companies launch entirely new models on the same day as each other. There's not a week that goes by that most of these platforms are not launching brand new models, brand new features. Everything is changing quickly enough that if you, as a brand, a seller, a marketer, any kind of company using AI, try to build an entire strategy around AI technology,
(22:59) that strategy will last for four to six weeks. Because in four to six weeks, there will be completely different technologies, new features that you didn't count on, that people have started to adopt, and adopted for five minutes, and then walked away from. If you don't believe that things can change really quickly, ask DeepSeek, whose usage chart went up and down like a classic bell
(23:22) curve.
**OpenAI's Changing Market Share**
(23:22) Ask OpenAI in a variety of ways. Sora 2 spiked and then settled down at a much lower rate. Ask them about ChatGPT, because one year ago, ChatGPT had 87% of global generative AI website traffic share. Again, this happens to be website data, not all data, because there are no sources that credibly combined the web share and the mobile share that I've
(23:48) seen at least. A year ago, ChatGPT had 87% of the share, and Gemini had 6%. In 12 months, ChatGPT lost a quarter of that share, and the Gemini share almost quadrupled, and every other platform combined roughly doubled in that time. Platform and feature adoption can change quickly. They are changing quickly, but the motivations for why people go and use
(24:16) these tools, I think, are going to be a lot more constant and evergreen. That's what we're working towards, and that's the model that we're going to start providing to brands, retailers, marketers, and sellers in the coming months.
**Building Blocks of AI Adoption Model**
(24:16) What I'm working on right now is building a data model to answer these questions in ways that brands can use. What we
(24:37) want to do is talk about what I'm calling the building blocks of AI adoption. This is something that is a work in progress. I'm going to publish this research in a couple of weeks. We're collecting the survey data to power this right now. What that means is I don't even know which order these different levels of behavior and motivation are going to end up in,
(24:57) because we're going to stack them in the order of prevalence that we see. The bottom of the building blocks is going to be total active adoption. Those are people who choose to use AI. It's not the 99% from that Gallup study at the start of the call who use AI without even realizing it. It's people choosing to use AI on a weekly basis. As we go up from there, we're going
**Motivation Categories: Asking & Doing**
(25:20) to see some non-mutually exclusive categories that refer to the "why," that talk about the motivations people have for using these technologies. One of the ones that we expect to be most common, based on all the data that I've seen collected everywhere else so far, is "asking." It's using AI to look for information or explanations. I'm going to show you five or six categories
(25:42) here. We're going to talk to brands and sellers about these five or six categories, but we're also going to dig deeper into the specific behaviors that add up to these categories. More than 50 different individual behaviors people are taking right now using artificial intelligence platforms. "Asking" is a pretty easy and broad category. It's people
(26:03) looking for facts and information and people looking for explanations. I think very likely the next building block up from that will be "doing," which is using AI for personal productivity, for advice, for guidance, and for tasks. There's a lot that can hide under here: writing or editing personal correspondence, creating and managing to-do lists, schedules,
(26:28) budgets, getting directions or navigations, translating from another language, brainstorming, as well as getting advice and guidance on a whole variety of different categories that we're surveying people on.
**Motivation Categories: Work, School & Shopping**
(26:28) How are people using AI to make their lives easier, more productive, more efficient on a personal level? They're not just using these tools at home.
(26:51) So we have a working building block that's going to get into using AI for work tasks and for school tasks. Then I'll be using AI for info and research, for correspondence, for images, audio, video, charts, graphs, presentation, for calculations and data analysis, computer coding, vibe coding, making apps, translations, but for work purposes, summarizing documents or meetings, and
(27:15) things like to-do lists and brainstorming. The one that I get the most questions about, but that I honestly think is not necessarily going to be that common or that big a building block, is shopping. Listen, I work at eMarketer. My clients are companies trying to sell
(27:39) things. They're very, very interested in this particular one. But I don't think that this is going to be huge. Daniel, it sounds I think you're saying you disagree that it's going to be huge. We can get into that afterwards, but the specific behaviors we're going to look at are using AI tools to get product recommendations, to look for information on specific products, comparing
(28:01) products, comparing prices, comparing stores, completing purchases on AI, or using AI for all this research and going somewhere else to complete that purchase.
**Motivation Categories: Relaxing & Connecting**
(28:01) We'll also be collecting it based on lots of different product categories. "Relaxing" is going to be a really interesting one: using AI for fun or to pass time, things like chatting
(28:21) with character AIs, creating or editing audio or video or images for fun, or writing text for fun, whether that's fiction or some other reason. Then perhaps the most interesting is "connecting." That's using AI as a friend, a companion, or a therapist. David can tell you all about his experience going on a date with an AI that would fit in the middle of these
(28:45) three bullets. Clearly, there are people using AI for companionship or for counseling or therapy. So this is what we're going to collect into these large categories you see as the building blocks, again, with more than 50 different specific behaviors that we'll be able to split the data on. Our hope is that these answers are going to help marketers and brands guide their AI
(29:08) strategy.
**Guiding AI Strategy: How, Why, Which Tools**
(29:08) How many of your customers use AI is the most important first question. That bottom building block, total active adoption, will show us the percentage of an audience that chooses to use AI each week. That'll define the importance of AI within a marketing plan. We're going to go with weekly. Of course, there's been a lot of discussion about OpenAI saying weekly
(29:27) data, sites weekly data. Google cites monthly data. I think if you're getting your haircut or paying your rent or mortgage more often than you're using ChatGPT, you're probably not really a ChatGPT user. So we want to look at this weekly and not monthly. Having said that, when we get into the specific granular, crunchy behaviors, like "did you compare prices?", we will look at
(29:50) that monthly because honestly, the weekly data, I think at this point, just won't turn up very much. It'll help us answer the question of why do our customers use AI? The motivations that are driving people into these technologies, and that will help us define the strategies. Because if it turns out your audience is using AI for relaxation and companionship, that's an incredibly different set of use cases than if your audience is mostly using it for looking for information and shopping.
(30:09) One of those instances very easily lends itself to marketing, advertising, and commercial purposes. The other one actually could lend itself to commercial purposes, but it's a much different scenario that requires a much different
(30:33) strategy to say, "Let's get a message through to people who are chatting with AI for fun or for companionship." That's going to be very different than putting something in front of someone who's using it basically as a search replacement or as Amazon replacement. Finally, it'll help us get to which AI tools our customers use. The
**Future Data & Open for Questions**
(30:52) platforms, the features, which of those 50 different behaviors are they engaging with? Which of the 10 leading AI platforms are they using? We'll have all that to offer brands and sellers to say, "This is where you execute. These are the tactics you use to execute that strategy." But while it's important, we do think it is the third question to ask, not the first question to ask,
(31:14) which is what we're so commonly seeing right now. That's where we are right now in our thinking. I'm really excited to get this data back from the field. We're working on collecting that data right now, and we'll be publishing all of this. I will post it in the AI Insiders AMG Slack. It'll be on my LinkedIn and everywhere else.
(31:38) There's a list of a bunch of stuff that we've been working on that I won't spend any time explaining to you because I've seen the chat lighting up. I've been trying to pay attention while presenting.
**Audience Questions: Shopping & Embedded AI**
(31:38) I'm really excited to hear people's thoughts and see where this lands for you guys.
(31:58) Loving this, Nate. It's so much fun to get this look even before it's all out there. Adam, go for it.
(31:58) Yeah. Nate, I was going to chime in on the shopping bit there and just generally wondering out loud if shopping is one of those things that people will make use of. But just the way I don't actively sit in front of a frontier platform and say, "Go to Amazon and tell me that people that
**AI Integration into UX**
(32:20) bought this may also like that," it's just embedded into the UX. Maybe shopping is the one more embedded, but it's a broader question: people have been asking this since the beginning, which is, what happens to the use of AI? People like David and myself are hacking away inside the base 44s, inside cloud code, and very active use of some kind of AI-
(32:48) first user experience journey. Whereas I would assume over time, this stuff just gets embedded into everything we're working on. There's no recommendation algorithm first UI interface at Goto anymore. Even in the early days of Google, I think a lot of people just started saying, "Wait a second, I'll just use Google's algorithm and I'll just put
(33:16) it directly on my site rather than `site:`."
**Meta AI & Google's Blurring Lines**
(33:16) Yeah. We're seeing some of that. Meta AI is an interesting example. Of course, they have their own model, but their definition of a Meta user is anyone who does any kind of search on Facebook, Instagram, or WhatsApp in a given month. My former
(33:41) boss referred to it as insidious. You can't avoid using Meta AI. If you need to find anything, congratulations, you're now counted as a Meta AI user. There are positives and negatives to that. If that's applied properly on a shopping site or a content site, then the AI can help you better find and more quickly get the
(34:04) information that you're looking for. It can get you where you're going faster. It will make it harder to count some of these behaviors. What we're collecting right now, we're looking at some of the largest embedded AIs within shopping sites: things like Amazon's Rufus and Walmart Sparky. There's no doubt that the companies involved in both the
(34:27) search space, the AI platform space, and the commerce space are all working as hard as they can to make my life as hard as possible when it comes to collecting this data. Google, in particular, has taken a particular joy in blurring the lines between traditional search and AI in a way that is very much to their benefit, that hopefully makes things much easier and more seamless for users, but
(34:51) that certainly makes it really clunky to think about the difference between a traditional search and an AI search when almost half of traditional searches throw up AI results. All of what you're saying makes sense to me. It's all stuff that we're thinking about, and we'll keep collecting that data as best we can. For now, this really interesting
(35:11) thing of people choosing to use AI, I think, is a really important distinction at this point in time that will become less important as time goes on, and we'll try to adapt to that as it happens.
**AEO vs. SEO Strategy**
(35:11) Who else has questions who can ask the analysts? This is a fun one we don't get every day. Hey, Nate. Thanks so much
(35:36) for giving this really informative. Here's a question. If I'm working with a client developing an AEO strategy for them, obviously, a lot of these sites are becoming invisible because of AI and make it searchable. Based on what you're finding, is there a parallel path, or do you want to almost start out first by answering those three
(35:58) questions about the user before you get into that, or do you see it as a parallel path?
(35:58) It's a great question, and I think there are two pieces to that. The first version of that that I hear pretty regularly is, how important is AEO or GEO compared to SEO?
**Is AI Killing Search? AEO vs. SEO Comparison**
(36:22) And the corollary question there is, "Is ChatGPT killing Google?" The second version of that question that I hear pretty regularly is, "How similar is AEO and GEO to SEO? Can I just do the same thing?" because a lot of very smart people, including Danny Sullivan, one of the OGs in search and SEO, are saying things like, "Good SEO
(36:49) is good GEO." So, I'll take those one by one. In terms of how important is AEO and GEO compared to SEO, as someone said earlier, no, in fact, AI is not killing search. GEO is not killing SEO. There are, again, certain categories and products where this is incredibly important already. I talk to B2B technology vendors, and this stuff is
(37:14) vitally important to how people are finding and evaluating their products. We've known that for a while. We've seen anecdotally a couple of years ago already, we were seeing really high ROI or really high conversion from traffic to warm lead between AI tools and these technology vendors, B2B technology vendors. I suspect that
(37:44) other very high consideration products, even on the consumer side, are experiencing something similar, although it's too early for us to have great data on that.
**Traditional Search Dominance (3.3% AI, 96.7% Traditional)**
(37:44) Overall, no, AI is not killing traditional search. I ran this data in September. I'm working to update it right now, now that it's been a few months. But in September of last year, I ran this
(38:04) analysis, and we combined the amount of time people spend looking for information. People spend in search-like behaviors on the top four AI platforms: ChatGPT, Gemini, Claude, and Perplexity were the four that we used in that case. We also looked at the amount of time people were spending on traditional search engines, and the only two we counted were Google
(38:26) and Bing. What we found was if you add up all the time people spent looking for things, searching basically on the top four AIs, it was 3.3% of the time they spent searching in total when you add up those AI platforms and Google and Bing. Of course, there are other places people search for things. I wasn't counting people searching for things in the Reddit
(38:50) toolbox or the YouTube toolbox, or any of those places. But these general search locations, traditional search engines and AI platforms, 3.3% of their time was spent searching for things in the AI, and 96.7% of the time was spent searching for things on traditional search. Although Bing has always been an industry punchline, it's a huge tool
(39:13) that makes a lot of money for Microsoft and drives a lot of value for a lot of advertisers, but it's so much smaller than Google search. It has been for years that we kind of joke about Bing. As of September of last year, people spent more time looking for things on Bing than on the top four AI tools combined. What that tells me is, yes, there are some
(39:36) products and categories and audiences for whom AI is a dominant and important way they're finding information and learning about products and services. The companies in those categories need to pay a lot of attention to AI and to GEO and AEO. But for most products, most categories, and certainly most audience segments out there, Google is still far and away the number one place people
(39:59) are going. When I hear marketers say, "We're just going to move all of our SEO initiative into GEO," for almost all of them, that would be a pretty big mistake.
**SEO Success ≠ GEO Success**
(40:08) The second part of that is, are AEO and GEO the same thing as SEO? Folks like Danny say good SEO is good GEO. I see people every day,
(40:23) I saw it again this morning, people trying to answer that question on a theoretical basis, using logic and the advice of experts.
(40:23) Sure, I love using logic and the advice of experts, but even more than that, I love using data. There's cold hard data that shows that success in SEO does not lead to success in GEO, and vice versa. Again, there are a million
(40:45) studies with a million different versions of this data. One I quote most commonly shows that eight or nine percent of the links that are cited in ChatGPT responses would show up in the first page of Google organic results for the same
(41:08) query, the same prompt run as a search query. Eight or nine percent from Gemini, eight or nine percent from ChatGPT, eight or nine percent from Co-pilot. It goes up a lot for Perplexity because they clearly are using a different concept for what they want to show. I've seen numbers as high as 30% and 40%. Those are definitely a lot better than 8% or 9%. But either way, the vast majority of the time that you have achieved success in SEO, landing on the
(41:33) first page of Google results, you have not achieved success in GEO. Whatever the theoretical, expert-driven opinions are on whether they should be the same thing, the reality is, when we look at the data, success in one does not lead to success in the other. That tells me that we need to think about different strategies and tactics. I like the dimensions and the
(41:56) perspective about consumers, customers, and industries. But I was wondering what lens we should put on for global brands around geographies, if any regions, and just some basic perspective on adoption across countries.
**Global AI Adoption & Local Data Importance**
(41:56) If it's hard to get this data across different behaviors and tools, it's even harder to get it across
(42:25) countries. We are constantly looking for data and trying to help our clients understand that. But again, we're in this situation where different studies have very different answers to that question.
(42:25) Yeah, that's fine. Totally. Trun, I think you're unmuted, and we can hear you.
(42:49) I did an analysis of the three largest studies I could find that included at least 20 different international markets adoption of AI, and I evaluated whether there were commonalities amongst their findings. None of those studies, if you categorized each of the countries into low, medium, or high AI adoption,
(43:17) none of them had more than 40% overlap with either of the others, which is infuriating and entirely typical of where we are with AI adoption data right now. So again, we're working on our own study to try to answer that question. But there's this other point: does it matter in Indonesia as much as elsewhere? I've seen data saying that
(43:43) Indonesia has surprisingly high AI adoption. I don't know a lot of marketers who are just targeting Indonesia. I've worked with Coca-Cola. They are most certainly targeting Indonesia as a whole place. Most brands, most companies are not just targeting the entire general population of a country. The question for me is a lot less about which
(44:04) countries have higher or lower adoption, and a lot more about what your target audience within each of your target countries is doing. For that reason, I don't think you need to get to a big, broad global study on it. I would say go and find reliable local data. I've seen fantastic local studies that look at six or seven different Latin American countries.
(44:28) I've seen, very bizarrely, the best study I found on Europe focuses on the Nordic region plus, I think, UK and Germany, and just seems to ignore Spain, France, and Italy. But go and find the local data and make sure you can cut it by the audience that you're looking for, and that will help you answer the first of those three questions I mentioned before.
**Recommended BTV Interview**
(44:52) All right, good one from Jim K. Let's see if Jim C could keep up and keep the bar high. Item number two. I just wanted if anyone wants to double-click on some of what you've said, Nate, they should go and watch your BTV interview of yesterday.
(44:52) Because it's what is it, eight or nine minutes long, but you do...
(45:13) much shorter than what I just dumped on you guys.
**Common Misconceptions & Data Quality in AI**
(45:13) It's all the search engines, and it's all the SEO, GEO. I'd watched it just before we started here, and it was very helpful for framing. That's it. I can't beat what the other Jim said.
(45:13) Other questions? I know Dan.
(45:13) Talk about Indonesia.
(45:35) We just want to talk about Indonesia. Daniel Green was hoping for some hot takes. Just a couple quick ones I'll borrow from him here. "Where does Nate think many people are wrong about AI today?" There are so many different things, so I won't say there's one thing that I think people are wrong about. I will say that most data I see is wrong. As I said, I've been doing this
(46:04) as an analyst for a quarter century now. I founded Jupiter Research's coverage of search marketing and led Forrester Research's coverage of social marketing, both at the periods of time when they were just starting to explode as marketing channels. In moments like the one we're in now with AI, in previous waves of technology, we've seen general confusion about what's
(46:31) happening, why it's happening, and how quickly it's happening. I have never seen anything like the amount of confusion and the amount of badly produced or incorrectly analyzed data that I see in AI. So, rather than answering that question directly, what I'll say is please, please, please check the base of the data that you're referencing, especially
(46:56) if it's survey data. The number of times I see people referencing data and saying X% of people think this about AI or do this with AI, and the base isn't general population. It's not online users. It is almost always things like active AI adopters, people who use AI tools at least once a week. I've seen data saying 80% of people trust AI responses more than organic search
(47:22) results. Well, 80% of people don't use AI.
**Critique of Synthetic Audiences**
(47:22) Nate, have you tried synthetic audiences and checked it against your own research to see how they compare?
(47:22) We are actively looking at using synthetic audiences. We don't think that they're ready. I have worked at other research firms not so long ago,
(47:47) where they were more bullish on synthetic audiences. My concern is this: it's the same as using AI to produce creative and things like that. These are probabilistic machines, which means they're averaging machines. They take the average of what's out there, and they give you something that makes sense
(48:10) within that context. That's not what's interesting about market research. If I wanted to make a pile of market research that looks like the last pile of market research I collected, a synthetic audience would do that for me every day. My challenge is how good are these things going to be at identifying and representing the outliers? I would say that about
(48:33) data collection and synthetic survey audiences. I would also say that about things like creative concepting, designing actual creative assets. The interesting things in those fields are not what looks most like the things that you can find that have gone before, but how do we break molds and patterns?
(48:56) But how about the marketers' perspective? Because if they're looking for user reactions, would they be able to get what to expect from the audience?
(48:56) Yes, it is by definition a backward-looking averaging machine. There are many good uses for it, and in most cases, I think it's probably perfectly fine to use it for
(49:28) synthetic audiences for survey work, for example. But again, most cases aren't the interesting ones. If I dig through fresh survey data, I'm actively looking for the pieces of data that stand out, not for the pieces of data that look like all the other data. And the piece of data that stands out, I then validate. I make sure that there isn't some mistake or
(49:56) other reason that it's standing out that it shouldn't, and make sure I asked the question properly and that the survey logic was working properly, and all of that. Once I know that this is a real result that stands out from the other results, then I start to look at what's happening and why. That's what's interesting. That's what turns into the reports that I write and
(50:15) the talks that I give. If you're just looking for the average of everything, if you're just looking for what the median person would say, and you create this audience that is a lot of people very close to that median person, it's a lot less interesting to me. I know that I'm oversimplifying enormously, and I know that there are potential ways of
(50:40) solving this problem. Personally, if they came to me and said, "We want to run your survey not with a real audience but with a synthetic audience," I would beg, borrow, and steal to avoid that happening right now.
**Concluding Remarks & Future Engagement**
(50:40) Well, Nate, we're going to have way more questions for you. Appreciate you coming
(50:58) on and answering quite a bunch, maybe some we didn't even know we had. As you have more to share, this is an open invite. I think it's a tremendous look that's not just, "Oh, here are some trends that are happening," but the stories behind them and what data you could trust, what you can't. It reminds me of some other areas I've
(51:23) seen research in, like research in millennials back in the day, where, "Okay, there are 80 million millennials." You can get the research to say whatever you want about them. They're both very into social causes and also very self-serving, just wanting to look out. You can find data to support anything with AI. See how it's so messy out there?
(51:45) Helping be a guide for what we can trust is eMarketer's heritage, it's your heritage. Glad there's a great fit together. Thank you for this, and I'm excited to keep getting to learn from you.
(51:58) Thanks so much. I'm happy to come back. As I said, when the data is ready, I will absolutely be posting it in your
(52:06) Slack and on various social channels, and I'm happy to come back and present it if that's of interest.
(52:06) Well, I think you'll get some takers here. No vetoes in this crowd, right? All right. Lots of applause going on. Thanks so much, and thanks everyone for always coming with your great questions and interest in the conversation today. I like how I looked
(52:24) away from the chat for a second, and it's like 20 new messages: "What the heck is going on here?" Nate, I could share that with you as well if you want to catch up on anything you missed. Thank you all. Next week we're off because I hope to see SEU and Architecture Live 3 in person, but then we'll be back with a great lineup of guest speakers coming up. See you all
(52:44) very soon.
(0:04) Hey everyone, I'm David Burkwitz. Welcome to another edition of AI Insiders with AI Marketers Guild from Architecture Media. I've got a guest; this one's been decades in the making. Nate Elliot is someone whose research, analysis, and insights I've been following for many years. We were, I guess, even quasi-competitors, or at least our firms
(0:32) were sometimes got along better than others and were navigating how to play well together. But he's someone I've been able to admire in that healthy competitive sense, from everything he's been doing, one of the ad and marketing industry's top analysts. Then once he joined my old home, eMarketer,
(0:58) where I spent a lot of my formative years in the industry really early on, and seeing what Nate is doing covering AI in certain areas as part of his purview. Nate was able to come to my office, which is the coffee shop in Koreatown, and we got to catch up on things. I'm just so excited to geek out here today and to get to have
(1:26) conversations with one of those people who's one of your gurus, who you actually get to meet and learn from. I'm like, Nate, as soon as you've got something ready to share, come on AMG. You're going to get some way better questions than I'll ask from this crew. If you haven't been here before, for everyone else, these are interactive conversations. Nate, I could go on for the whole hour welcoming you, but that wouldn't be fair to our crew here. Welcome.
(1:47) Thanks very much. Earl, I think you were clapping, David, saying you were going to stop. Is that what you're...
(2:08) Oh, yeah. That was me. That was me clapping. I will neither confirm nor deny. Excellent. Well, thank you, David, for having me. I'm excited to be here. Thanks to everyone else for joining today, both live and anyone who watches the recording. Hello, future people.
**Nate Elliott's Background & Focus**
(2:19) My name is Nate Elliot. I am a principal analyst at eMarketer. I cover how AI is changing marketing, commerce,
(2:27) and the customer journey. As David said, I've been an analyst in the industry for quite some time. I actually worked at DoubleClick back in the 90s when we were helping to invent the way advertising would work, for better and for worse. I joined Jupiter Research as an analyst in 2003, and since then I've been an analyst ever since, at Jupiter, then Forrester. I ran my own shop for
(2:48) about a decade, and I was thrilled to join eMarketer last summer and get to work as their lead analyst on AI. What I want to share with you guys, and forgive me, we're a Google shop, not a Zoom shop, so I'm going to ask you, David, to tell me if I'm getting this right as I share my slides. I still don't know how to keep an eye on the Zoom chat while
(3:12) I'm presenting. So, I'll surface if someone's shouting things out. What I want to talk about is the building blocks of AI adoption. I've been looking at this space even before I came to eMarketer to lead the coverage of AI.
**The Quest for AI Adoption Truth**
(3:20) Of course, I was paying attention. I was running surveys and research on this for companies, including Walmart and
(3:34) Burger King, and a bunch of others. One of my first tasks when I got to eMarketer was to create our single version of the truth around what's happening in AI adoption. All the stuff we want to talk about, the analysis we want to provide to platforms, to vendors, to brands, be they marketers or sellers, all of that is analytical. I would dare
(3:56) say there is no truth. There are simply informed opinions, and that's where the job gets really interesting. As a baseline, one of the things I wanted to understand was how many people are actually using these tools? Because it's a thing we talk about all the time. It's a concept, a number that I honestly don't believe the industry has put really good
(4:20) analysis or factual data around.
**The Discrepancy in AI Usage Data**
(4:20) I see numbers that are absolutely all over the place. One of the things that I like to do when I'm in a room with people is ask everyone to shout out what percentage of the population uses AI in the US. I did this at a couple events last week. I got numbers as low as below 20% and as high as 85 or 90%. The joke, of
(4:46) course, is that whatever number you say, as long as you're between, say, 10% and 99%, you're correct because I can find data from a credible source that will back you up, that will say that yes, that is the number of people who are using it. To answer the question today, I want to use a piece of research that was published a little over a year
**Gallup Study: AI Usage Awareness vs. Reality**
(5:00) ago. It was conducted in December of 2024, and I'm fully aware of it. Showing you AI usage data from 2024 is like showing you space travel data from 1944. It doesn't make a lot of sense, but hopefully it will in just a moment. This is a study that Gallup published last January 2025, that ran in December of 2024, and they asked about 4,000 US online users, "Do you recall
(5:39) using an AI-enabled product in the past 7 days?" They weren't just asking about ChatGPT and Gemini and things like that. They were saying, "Is there anything else you've used that is AI-powered or AI-enabled?" And 36% of US online users said that they could recall having done that in the past week. Then Gallup came back and said, "While you're here, let's just ask if you used
(6:01) any of the following tools."
**99% Use AI (Known or Unknown)**
(6:01) It turned out that 99% of these same exact respondents had in fact used one of these AI-enabled technologies in the previous seven days. Everyone uses AI at this point,
(6:26) whether they know it or not. That's been true, of course, for at least a year or 15 months, probably longer than that. I'm sure that I'm preaching to the choir a little bit. But one of the reasons I want to show this, even to an educated and interested group like the AI Insiders, is to reinforce this gap between people choosing to use AI, or being aware of using AI, and actually
(6:50) using or interacting with these technologies. That gap makes a huge difference to the brands, the marketers, and the sellers who are looking at using AI as a marketing channel. Helping those brands, marketers, and sellers figure out how they can best leverage AI as a marketing channel is a big part of my remit. It's a big part of the work that I'm doing.
(7:12) The thing is, not all AI adoption is created equal. While lots of people use AI, 99% of people in that study from a year and three months ago had actually interacted with an AI tool in the previous month.
**Adoption vs. Adaptation in AI Usage**
(7:12) Not all of them had chosen to do so. Even amongst those who choose to use AI, the adoption is really high, but the adaptation, as my colleague Jacob Bourne
(7:37) would call it, is really low. What we talk about with adoption versus adaptation is adoption are people who use the tools. They go and they choose to take advantage of the tools. Adaptation, we're talking more about changing behavior patterns: taking things you've done for years online and moving those behaviors from one tool or platform or location to another tool or
(8:00) platform or location. One of the big behavior patterns that we see, of course, is people going to traditional search engines, Google. The number one reason people go online is to find information. That's been true for a couple of decades now. Google has, depending on the market you're looking at, anywhere from 70 to 95% market share in traditional search.
(8:21) When we look at traditional search engines, we know that 95% of people who go online every month are using a traditional search engine at least once a month. And 86% of online users, almost all the people who go to a search engine monthly, are going there on a frequent basis, 10 or more times per month, at least about once every third day.
**AI Tool Adoption: A Sampling Phase**
(8:42) When we look at top AI tools, the numbers look a little bit different. It's much earlier in the adoption curve. So the total adoption, the total number of monthly users, is not going to be as high. 38% as of August of last year is still a remarkable number for a set of tools that at that point were less than three years old as a category. What's really interesting to me is that
(9:05) only about half of that 38% were choosing to go to these tools on a very regular basis, 10 or more times per month. A lot of people are playing with AI. They're sampling AI. They're using it unintentionally, or they're using it intentionally every now and then. While we all live in this bubble where everyone uses AI all the time, if you look at the general online
(9:29) population, the reality is a lot of people use AI a little bit, and a handful of people are using AI a lot. That distinction, the gap between those polls, is really interesting and important, I think, to the marketers and sellers who are looking at using AI as a channel.
**Is AI Mobile-First?**
(9:48) Nate, quick question here.
(9:51) I mean, AI just by its nature also strikes me as something that is so mobile-first. Do some reports like this give short shrift to that concept, that you want AI to be with you everywhere?
(9:51) I'm sorry to cut you off. I wouldn't say it's mobile-first. It's snackable interactions,
(10:15) and that's often on mobile. I do a lot of my stuff on desktop.
(10:15) Oh, I do too, but I feel for most who are creating a recipe or telling a kid's story...
(10:15) That's a fair point. You're right.
(10:15) No, but both can be true.
(10:34) Mobile's a huge part of the story, if for no other reason than mobile's a huge
**Challenges in AI Usage Data Collection**
(10:34) part of the story of how people use all digital technology at this point. The study that I'm showing you happens to be based on desktop data analysis and clickstream data. Again, it gets to my point that it's really hard to find a good single source of truth around how many people are using these tools, which tools they're using, how frequently they're using them, and for exactly what
(10:53) purposes. There are imperfections in methodology no matter what study you look at. There is no research methodology that's going to perfectly give us this data. This is an imperfect study alongside a lot of other imperfect studies. I will soon be publishing my own imperfect studies, but hopefully in a way that add
(11:15) to the clickstream observational data, like what you see on the screen right now, to add some nuance to the conversation and to fill in some really important missing gaps that we see not just in studies like this, because Semrush and Dallas did a great job on this research, but it's limited by the data with which they were working. So I want to start to fill in those
(11:38) gaps with a consumer survey methodology, and we'll talk about that in a moment.
**OpenAI's Data Dump & Challenges for Marketers**
(11:38) People use AI in lots of different ways. This is great data. Full credit to OpenAI. The study they published called "How People Use ChatGPT" in September of last year is one of the greatest corporate data dumps I've ever
(12:02) seen in a quarter century of being an industry analyst. They update a lot of it, not all of it, but a lot of it on an ongoing basis for us, which is incredibly generous, and is one of the handful of things they do that still reference their founding as OpenAI, a company that would share its knowledge and information with the world. I show this
(12:23) slide because this is really hard to process. It's, I think, seven different primary categories, two dozen different subcategories, and there's this enormous range of different ways in which people are using artificial intelligence. Brands and retailers look at this, and they come to companies like eMarketer, and they say, "What am
(12:47) I supposed to do with this? Where am I actually supposed to start if I'm hoping to reach people who are using these channels?" I'll get it out of the way upfront. There's an entire section of this industry that is not using AI as a channel, but is using AI to make your work better as a marketer, a seller, a brand, a retailer. That is research that
(13:07) we are working on and will be producing. But a lot of what I'm talking about today is how people are using AI, and therefore, how brands and sellers can use it as a channel to reach those people. When we talk to brands and sellers, we think, rather than look at this enormous amount of data that's out there, rather than look at slides like that colorful spaghetti I
(13:26) just showed you on the previous chart, ask these three questions, and ask them in this order. First and foremost, how many of our customers use AI?
**Customer-Specific AI Usage**
(13:26) That's a more difficult question than a lot of us probably understand because in this little bubble that we all work in as AI Insiders, and people who think a lot about this, we assume that the vast
(13:49) majority of the world uses AI. You hear things like AI is replacing Google. I've seen people stand in front of slides that say 80% of the population goes and uses LLMs every month, and I've never seen really credible data that says that's true. But I have seen data that says that's true. Whatever the overall number is, I think the first thing that brands
(14:13) and sellers have to ask themselves is not how much of the general population uses AI, but how many of your customers use AI?
**Generational AI Usage: Most Adoptive**
(14:13) I want to play a game with you guys. I'll ask you to pop it in the chat or shout it out, but which of these generations do you think uses AI the most in the US right now? Any guesses? David, what's your guess?
(14:35) I want to say Gen X, but I feel millennials are still going to be a little more tech-forward. So, I'll say millennials.
(14:35) We'll say millennials. I see people go in the chat. I'm seeing a bunch of different guesses in there. Gen X, Millennials, a lot of people are focusing on Gen X and millennials. Those are some solid
(14:58) guesses. The actual answer is category D, Gen Z.
**Generational AI Usage: Least Adoptive**
(15:00) Let's flip that coin around though. That's the population, the generation that uses AI the most as a percentage of online users. Which generation uses AI the least? David, let's hear your guess. I know you have a guess on this.
(15:00) Well, Gen Alpha.
(15:00) You think Gen Alpha?
(15:20) Yeah.
(15:20) All right, let's see what else is going on in here. An even mix of boomers and Gen Alpha. A couple people said maybe it's Generation X. The generation that uses AI the least, you guys have gotten this right. Congratulations, more than any other audience I've asked this question. It's
(15:42) actually Gen Alpha. The reason is this: a lot of Gen Alpha is online. We run these numbers in our forecasts as a percentage of online users. That's what I'm looking at here. Something like three-quarters of Gen Alpha is online. That's terrifying to me because Gen Alpha right now is between the ages of two and 13 years old. For 75% of them to
(16:06) be online is scary. Our forecast team assures me that if a parent starts YouTube on their tablet and hands it to the toddler in the stroller, that is in fact an online individual. That's how we get to 75%. The reason that those online users in Gen Alpha aren't using AI is that most of them don't know how to read and write.
**Gen Alpha's AI Adoption & Future Projections**
(16:33) I do wonder, and this is my own bias here, having a 12-year-old, that my 12-year-old is so skeptical herself. I feel there's, and we often see this, generations back-to-back often one rebels against the other. So if my daughter sees all these college kids and 20-somethings obsessed with AI, it's like, screw them, right?
(16:56) Yeah. There are definitely a lot of motivations that we didn't get into because, to answer Selena's question, this isn't a study, it's a forecast. This is a model that was put together by our forecasting team using thousands of different data points from hundreds of different sources, then modeling it forward based on how we see the market
(17:16) changing over time. We do think that once these two to 13-year-olds—I'm assuming, David, that your 12-year-old is very good at reading and writing, but the 2-year-olds certainly don't know how to read and write. The average age at which someone in the US learns to read and write basic words is 6 years old, I believe. That means over the next few years, as Gen Alpha gets more
(17:38) literate, they will start to use AI more, and we're forecasting that they'll overtake boomers in 2027 by the end of next year.
**Gen Z Leads AI Adoption, Challenging Assumptions**
(17:38) Anyone who guessed millennials as the biggest adopters of AI, you would have been right about a year and a half ago. But according to our model, Gen Z overtook millennials towards the end of 2025. The reason I love your
(18:04) comment, David, your 12-year-old called Crusty, that sounds right, not because of you, but because of Gen Alphas. The reason I show this and the reason I play this game, which hopefully you guys found fun, maybe enlightening, is to show you that we are some of the most interested people about AI adoption and how AI is changing our world and our
(18:29) industry. A lot of us got this wrong. I would have gotten this wrong if you'd asked me this question before I sat down with the eMarketer forecast on this topic. We make a lot of assumptions, and a lot of those assumptions are based on what the people directly around us are doing. The people around us are not the general population, and they certainly don't, on average,
(18:50) reflect the entire spread of individual groups, be they generational, gender, race, income, education, whatever it is. The people immediately around us don't often give us the best lens to see what's really happening in the world. That's why we think the first question that is really important for marketers and sellers to get great data on is what percentage of your audience
(19:12) specifically is using AI, because that will start to tell you how important AI is in your marketing and your sales strategy.
**AI's Role: Primary or Secondary Channel?**
(19:12) I don't think there's a seller or a brand, no matter their audience, that shouldn't be working with AI. But I do know that there are some sellers or brands for whom AI is perhaps the single most important channel for getting messages out and marketing to and
(19:37) selling to the people that they're trying to reach. And I know that there are a lot of brands, probably many more, for whom AI should remain at this point a secondary channel that they're leveraging to a limited extent, while keeping their focus on the larger, more important channels today, and also testing and learning and preparing themselves for when that tipping point
(19:59) eventually comes. It's a really important distinction because I've talked to marketers, I've talked to CMOs who think we need to drop everything and only focus on AI right now. There are a handful of companies and product categories in which that is true, but that is not what is happening for the vast majority of brands, fans, marketers, and sellers
(20:22) right now.
**Understanding Customer Motivation for AI Use**
(20:22) I think understanding exactly how much your audience is choosing to adopt these tools is a really important starting place that a lot of the people I talk to aren't actually starting their investigations. Once we answer that question, I think the next question we have to answer is, why do our customers use AI? When I talk about why, I want to
(20:43) be clear. I'm not talking about what features or tools they're going to these platforms to use. Not, "Hey, the why is they want to play with Sora 2," or, "the why is they want to play with a character AI." Those are important things. We're going to get to that in the third question, but I want to know why people are actually choosing these tools in the first place. Why are
(21:05) they going to these platforms? What is the human motivation that drives them there?
**SEO is Not Dead & Mixed Data Challenges**
(21:05) To Adam's question in the chat, no, SEO is nowhere near dead. We can talk about that in a little while. But when we talk about why, most of the data I see, and again, I'm having a really hard time finding single sources of truth on this, most of the AI usage data I see, whether it's
(21:26) survey-based, whether it is behavioral and observational, at the very best mixes motivations and features. That OpenAI slide, that rainbow spaghetti I showed you a few minutes ago, talks about why. And Daniel Green, yes, there are some Anthropic and OpenAI reports on this. But just like the AP survey on the screen right now, just like the OpenAI data I
(21:50) showed you a few minutes ago, what we're really seeing is a mix of motivations and also the technologies, the features, the actual tools that they're using. I don't think the tools are as important a question as the features. There are a couple of reasons for that, but let me skip ahead and get to that.
(22:13) The third thing is, which tools do our customers use?
**AI Tools & Features: Rapidly Changing Landscape**
(22:13) First, how many of our customers are using AI? Second, what is their motivation for going and using these tools? And the third question is, which tools do they use? The reason I don't think the tools and the features are as important is because they're changing every day. There are days in
(22:32) which multiple frontier platform companies launch entirely new models on the same day as each other. There's not a week that goes by that most of these platforms are not launching brand new models, brand new features. Everything is changing quickly enough that if you, as a brand, a seller, a marketer, any kind of company using AI, try to build an entire strategy around AI technology,
(22:59) that strategy will last for four to six weeks. Because in four to six weeks, there will be completely different technologies, new features that you didn't count on, that people have started to adopt, and adopted for five minutes, and then walked away from. If you don't believe that things can change really quickly, ask DeepSeek, whose usage chart went up and down like a classic bell
(23:22) curve.
**OpenAI's Changing Market Share**
(23:22) Ask OpenAI in a variety of ways. Sora 2 spiked and then settled down at a much lower rate. Ask them about ChatGPT, because one year ago, ChatGPT had 87% of global generative AI website traffic share. Again, this happens to be website data, not all data, because there are no sources that credibly combined the web share and the mobile share that I've
(23:48) seen at least. A year ago, ChatGPT had 87% of the share, and Gemini had 6%. In 12 months, ChatGPT lost a quarter of that share, and the Gemini share almost quadrupled, and every other platform combined roughly doubled in that time. Platform and feature adoption can change quickly. They are changing quickly, but the motivations for why people go and use
(24:16) these tools, I think, are going to be a lot more constant and evergreen. That's what we're working towards, and that's the model that we're going to start providing to brands, retailers, marketers, and sellers in the coming months.
**Building Blocks of AI Adoption Model**
(24:16) What I'm working on right now is building a data model to answer these questions in ways that brands can use. What we
(24:37) want to do is talk about what I'm calling the building blocks of AI adoption. This is something that is a work in progress. I'm going to publish this research in a couple of weeks. We're collecting the survey data to power this right now. What that means is I don't even know which order these different levels of behavior and motivation are going to end up in,
(24:57) because we're going to stack them in the order of prevalence that we see. The bottom of the building blocks is going to be total active adoption. Those are people who choose to use AI. It's not the 99% from that Gallup study at the start of the call who use AI without even realizing it. It's people choosing to use AI on a weekly basis. As we go up from there, we're going
**Motivation Categories: Asking & Doing**
(25:20) to see some non-mutually exclusive categories that refer to the "why," that talk about the motivations people have for using these technologies. One of the ones that we expect to be most common, based on all the data that I've seen collected everywhere else so far, is "asking." It's using AI to look for information or explanations. I'm going to show you five or six categories
(25:42) here. We're going to talk to brands and sellers about these five or six categories, but we're also going to dig deeper into the specific behaviors that add up to these categories. More than 50 different individual behaviors people are taking right now using artificial intelligence platforms. "Asking" is a pretty easy and broad category. It's people
(26:03) looking for facts and information and people looking for explanations. I think very likely the next building block up from that will be "doing," which is using AI for personal productivity, for advice, for guidance, and for tasks. There's a lot that can hide under here: writing or editing personal correspondence, creating and managing to-do lists, schedules,
(26:28) budgets, getting directions or navigations, translating from another language, brainstorming, as well as getting advice and guidance on a whole variety of different categories that we're surveying people on.
**Motivation Categories: Work, School & Shopping**
(26:28) How are people using AI to make their lives easier, more productive, more efficient on a personal level? They're not just using these tools at home.
(26:51) So we have a working building block that's going to get into using AI for work tasks and for school tasks. Then I'll be using AI for info and research, for correspondence, for images, audio, video, charts, graphs, presentation, for calculations and data analysis, computer coding, vibe coding, making apps, translations, but for work purposes, summarizing documents or meetings, and
(27:15) things like to-do lists and brainstorming. The one that I get the most questions about, but that I honestly think is not necessarily going to be that common or that big a building block, is shopping. Listen, I work at eMarketer. My clients are companies trying to sell
(27:39) things. They're very, very interested in this particular one. But I don't think that this is going to be huge. Daniel, it sounds I think you're saying you disagree that it's going to be huge. We can get into that afterwards, but the specific behaviors we're going to look at are using AI tools to get product recommendations, to look for information on specific products, comparing
(28:01) products, comparing prices, comparing stores, completing purchases on AI, or using AI for all this research and going somewhere else to complete that purchase.
**Motivation Categories: Relaxing & Connecting**
(28:01) We'll also be collecting it based on lots of different product categories. "Relaxing" is going to be a really interesting one: using AI for fun or to pass time, things like chatting
(28:21) with character AIs, creating or editing audio or video or images for fun, or writing text for fun, whether that's fiction or some other reason. Then perhaps the most interesting is "connecting." That's using AI as a friend, a companion, or a therapist. David can tell you all about his experience going on a date with an AI that would fit in the middle of these
(28:45) three bullets. Clearly, there are people using AI for companionship or for counseling or therapy. So this is what we're going to collect into these large categories you see as the building blocks, again, with more than 50 different specific behaviors that we'll be able to split the data on. Our hope is that these answers are going to help marketers and brands guide their AI
(29:08) strategy.
**Guiding AI Strategy: How, Why, Which Tools**
(29:08) How many of your customers use AI is the most important first question. That bottom building block, total active adoption, will show us the percentage of an audience that chooses to use AI each week. That'll define the importance of AI within a marketing plan. We're going to go with weekly. Of course, there's been a lot of discussion about OpenAI saying weekly
(29:27) data, sites weekly data. Google cites monthly data. I think if you're getting your haircut or paying your rent or mortgage more often than you're using ChatGPT, you're probably not really a ChatGPT user. So we want to look at this weekly and not monthly. Having said that, when we get into the specific granular, crunchy behaviors, like "did you compare prices?", we will look at
(29:50) that monthly because honestly, the weekly data, I think at this point, just won't turn up very much. It'll help us answer the question of why do our customers use AI? The motivations that are driving people into these technologies, and that will help us define the strategies. Because if it turns out your audience is using AI for relaxation and companionship, that's an incredibly different set of use cases than if your audience is mostly using it for looking for information and shopping.
(30:09) One of those instances very easily lends itself to marketing, advertising, and commercial purposes. The other one actually could lend itself to commercial purposes, but it's a much different scenario that requires a much different
(30:33) strategy to say, "Let's get a message through to people who are chatting with AI for fun or for companionship." That's going to be very different than putting something in front of someone who's using it basically as a search replacement or as Amazon replacement. Finally, it'll help us get to which AI tools our customers use. The
**Future Data & Open for Questions**
(30:52) platforms, the features, which of those 50 different behaviors are they engaging with? Which of the 10 leading AI platforms are they using? We'll have all that to offer brands and sellers to say, "This is where you execute. These are the tactics you use to execute that strategy." But while it's important, we do think it is the third question to ask, not the first question to ask,
(31:14) which is what we're so commonly seeing right now. That's where we are right now in our thinking. I'm really excited to get this data back from the field. We're working on collecting that data right now, and we'll be publishing all of this. I will post it in the AI Insiders AMG Slack. It'll be on my LinkedIn and everywhere else.
(31:38) There's a list of a bunch of stuff that we've been working on that I won't spend any time explaining to you because I've seen the chat lighting up. I've been trying to pay attention while presenting.
**Audience Questions: Shopping & Embedded AI**
(31:38) I'm really excited to hear people's thoughts and see where this lands for you guys.
(31:58) Loving this, Nate. It's so much fun to get this look even before it's all out there. Adam, go for it.
(31:58) Yeah. Nate, I was going to chime in on the shopping bit there and just generally wondering out loud if shopping is one of those things that people will make use of. But just the way I don't actively sit in front of a frontier platform and say, "Go to Amazon and tell me that people that
**AI Integration into UX**
(32:20) bought this may also like that," it's just embedded into the UX. Maybe shopping is the one more embedded, but it's a broader question: people have been asking this since the beginning, which is, what happens to the use of AI? People like David and myself are hacking away inside the base 44s, inside cloud code, and very active use of some kind of AI-
(32:48) first user experience journey. Whereas I would assume over time, this stuff just gets embedded into everything we're working on. There's no recommendation algorithm first UI interface at Goto anymore. Even in the early days of Google, I think a lot of people just started saying, "Wait a second, I'll just use Google's algorithm and I'll just put
(33:16) it directly on my site rather than `site:`."
**Meta AI & Google's Blurring Lines**
(33:16) Yeah. We're seeing some of that. Meta AI is an interesting example. Of course, they have their own model, but their definition of a Meta user is anyone who does any kind of search on Facebook, Instagram, or WhatsApp in a given month. My former
(33:41) boss referred to it as insidious. You can't avoid using Meta AI. If you need to find anything, congratulations, you're now counted as a Meta AI user. There are positives and negatives to that. If that's applied properly on a shopping site or a content site, then the AI can help you better find and more quickly get the
(34:04) information that you're looking for. It can get you where you're going faster. It will make it harder to count some of these behaviors. What we're collecting right now, we're looking at some of the largest embedded AIs within shopping sites: things like Amazon's Rufus and Walmart Sparky. There's no doubt that the companies involved in both the
(34:27) search space, the AI platform space, and the commerce space are all working as hard as they can to make my life as hard as possible when it comes to collecting this data. Google, in particular, has taken a particular joy in blurring the lines between traditional search and AI in a way that is very much to their benefit, that hopefully makes things much easier and more seamless for users, but
(34:51) that certainly makes it really clunky to think about the difference between a traditional search and an AI search when almost half of traditional searches throw up AI results. All of what you're saying makes sense to me. It's all stuff that we're thinking about, and we'll keep collecting that data as best we can. For now, this really interesting
(35:11) thing of people choosing to use AI, I think, is a really important distinction at this point in time that will become less important as time goes on, and we'll try to adapt to that as it happens.
**AEO vs. SEO Strategy**
(35:11) Who else has questions who can ask the analysts? This is a fun one we don't get every day. Hey, Nate. Thanks so much
(35:36) for giving this really informative. Here's a question. If I'm working with a client developing an AEO strategy for them, obviously, a lot of these sites are becoming invisible because of AI and make it searchable. Based on what you're finding, is there a parallel path, or do you want to almost start out first by answering those three
(35:58) questions about the user before you get into that, or do you see it as a parallel path?
(35:58) It's a great question, and I think there are two pieces to that. The first version of that that I hear pretty regularly is, how important is AEO or GEO compared to SEO?
**Is AI Killing Search? AEO vs. SEO Comparison**
(36:22) And the corollary question there is, "Is ChatGPT killing Google?" The second version of that question that I hear pretty regularly is, "How similar is AEO and GEO to SEO? Can I just do the same thing?" because a lot of very smart people, including Danny Sullivan, one of the OGs in search and SEO, are saying things like, "Good SEO
(36:49) is good GEO." So, I'll take those one by one. In terms of how important is AEO and GEO compared to SEO, as someone said earlier, no, in fact, AI is not killing search. GEO is not killing SEO. There are, again, certain categories and products where this is incredibly important already. I talk to B2B technology vendors, and this stuff is
(37:14) vitally important to how people are finding and evaluating their products. We've known that for a while. We've seen anecdotally a couple of years ago already, we were seeing really high ROI or really high conversion from traffic to warm lead between AI tools and these technology vendors, B2B technology vendors. I suspect that
(37:44) other very high consideration products, even on the consumer side, are experiencing something similar, although it's too early for us to have great data on that.
**Traditional Search Dominance (3.3% AI, 96.7% Traditional)**
(37:44) Overall, no, AI is not killing traditional search. I ran this data in September. I'm working to update it right now, now that it's been a few months. But in September of last year, I ran this
(38:04) analysis, and we combined the amount of time people spend looking for information. People spend in search-like behaviors on the top four AI platforms: ChatGPT, Gemini, Claude, and Perplexity were the four that we used in that case. We also looked at the amount of time people were spending on traditional search engines, and the only two we counted were Google
(38:26) and Bing. What we found was if you add up all the time people spent looking for things, searching basically on the top four AIs, it was 3.3% of the time they spent searching in total when you add up those AI platforms and Google and Bing. Of course, there are other places people search for things. I wasn't counting people searching for things in the Reddit
(38:50) toolbox or the YouTube toolbox, or any of those places. But these general search locations, traditional search engines and AI platforms, 3.3% of their time was spent searching for things in the AI, and 96.7% of the time was spent searching for things on traditional search. Although Bing has always been an industry punchline, it's a huge tool
(39:13) that makes a lot of money for Microsoft and drives a lot of value for a lot of advertisers, but it's so much smaller than Google search. It has been for years that we kind of joke about Bing. As of September of last year, people spent more time looking for things on Bing than on the top four AI tools combined. What that tells me is, yes, there are some
(39:36) products and categories and audiences for whom AI is a dominant and important way they're finding information and learning about products and services. The companies in those categories need to pay a lot of attention to AI and to GEO and AEO. But for most products, most categories, and certainly most audience segments out there, Google is still far and away the number one place people
(39:59) are going. When I hear marketers say, "We're just going to move all of our SEO initiative into GEO," for almost all of them, that would be a pretty big mistake.
**SEO Success ≠ GEO Success**
(40:08) The second part of that is, are AEO and GEO the same thing as SEO? Folks like Danny say good SEO is good GEO. I see people every day,
(40:23) I saw it again this morning, people trying to answer that question on a theoretical basis, using logic and the advice of experts.
(40:23) Sure, I love using logic and the advice of experts, but even more than that, I love using data. There's cold hard data that shows that success in SEO does not lead to success in GEO, and vice versa. Again, there are a million
(40:45) studies with a million different versions of this data. One I quote most commonly shows that eight or nine percent of the links that are cited in ChatGPT responses would show up in the first page of Google organic results for the same
(41:08) query, the same prompt run as a search query. Eight or nine percent from Gemini, eight or nine percent from ChatGPT, eight or nine percent from Co-pilot. It goes up a lot for Perplexity because they clearly are using a different concept for what they want to show. I've seen numbers as high as 30% and 40%. Those are definitely a lot better than 8% or 9%. But either way, the vast majority of the time that you have achieved success in SEO, landing on the
(41:33) first page of Google results, you have not achieved success in GEO. Whatever the theoretical, expert-driven opinions are on whether they should be the same thing, the reality is, when we look at the data, success in one does not lead to success in the other. That tells me that we need to think about different strategies and tactics. I like the dimensions and the
(41:56) perspective about consumers, customers, and industries. But I was wondering what lens we should put on for global brands around geographies, if any regions, and just some basic perspective on adoption across countries.
**Global AI Adoption & Local Data Importance**
(41:56) If it's hard to get this data across different behaviors and tools, it's even harder to get it across
(42:25) countries. We are constantly looking for data and trying to help our clients understand that. But again, we're in this situation where different studies have very different answers to that question.
(42:25) Yeah, that's fine. Totally. Trun, I think you're unmuted, and we can hear you.
(42:49) I did an analysis of the three largest studies I could find that included at least 20 different international markets adoption of AI, and I evaluated whether there were commonalities amongst their findings. None of those studies, if you categorized each of the countries into low, medium, or high AI adoption,
(43:17) none of them had more than 40% overlap with either of the others, which is infuriating and entirely typical of where we are with AI adoption data right now. So again, we're working on our own study to try to answer that question. But there's this other point: does it matter in Indonesia as much as elsewhere? I've seen data saying that
(43:43) Indonesia has surprisingly high AI adoption. I don't know a lot of marketers who are just targeting Indonesia. I've worked with Coca-Cola. They are most certainly targeting Indonesia as a whole place. Most brands, most companies are not just targeting the entire general population of a country. The question for me is a lot less about which
(44:04) countries have higher or lower adoption, and a lot more about what your target audience within each of your target countries is doing. For that reason, I don't think you need to get to a big, broad global study on it. I would say go and find reliable local data. I've seen fantastic local studies that look at six or seven different Latin American countries.
(44:28) I've seen, very bizarrely, the best study I found on Europe focuses on the Nordic region plus, I think, UK and Germany, and just seems to ignore Spain, France, and Italy. But go and find the local data and make sure you can cut it by the audience that you're looking for, and that will help you answer the first of those three questions I mentioned before.
**Recommended BTV Interview**
(44:52) All right, good one from Jim K. Let's see if Jim C could keep up and keep the bar high. Item number two. I just wanted if anyone wants to double-click on some of what you've said, Nate, they should go and watch your BTV interview of yesterday.
(44:52) Because it's what is it, eight or nine minutes long, but you do...
(45:13) much shorter than what I just dumped on you guys.
**Common Misconceptions & Data Quality in AI**
(45:13) It's all the search engines, and it's all the SEO, GEO. I'd watched it just before we started here, and it was very helpful for framing. That's it. I can't beat what the other Jim said.
(45:13) Other questions? I know Dan.
(45:13) Talk about Indonesia.
(45:35) We just want to talk about Indonesia. Daniel Green was hoping for some hot takes. Just a couple quick ones I'll borrow from him here. "Where does Nate think many people are wrong about AI today?" There are so many different things, so I won't say there's one thing that I think people are wrong about. I will say that most data I see is wrong. As I said, I've been doing this
(46:04) as an analyst for a quarter century now. I founded Jupiter Research's coverage of search marketing and led Forrester Research's coverage of social marketing, both at the periods of time when they were just starting to explode as marketing channels. In moments like the one we're in now with AI, in previous waves of technology, we've seen general confusion about what's
(46:31) happening, why it's happening, and how quickly it's happening. I have never seen anything like the amount of confusion and the amount of badly produced or incorrectly analyzed data that I see in AI. So, rather than answering that question directly, what I'll say is please, please, please check the base of the data that you're referencing, especially
(46:56) if it's survey data. The number of times I see people referencing data and saying X% of people think this about AI or do this with AI, and the base isn't general population. It's not online users. It is almost always things like active AI adopters, people who use AI tools at least once a week. I've seen data saying 80% of people trust AI responses more than organic search
(47:22) results. Well, 80% of people don't use AI.
**Critique of Synthetic Audiences**
(47:22) Nate, have you tried synthetic audiences and checked it against your own research to see how they compare?
(47:22) We are actively looking at using synthetic audiences. We don't think that they're ready. I have worked at other research firms not so long ago,
(47:47) where they were more bullish on synthetic audiences. My concern is this: it's the same as using AI to produce creative and things like that. These are probabilistic machines, which means they're averaging machines. They take the average of what's out there, and they give you something that makes sense
(48:10) within that context. That's not what's interesting about market research. If I wanted to make a pile of market research that looks like the last pile of market research I collected, a synthetic audience would do that for me every day. My challenge is how good are these things going to be at identifying and representing the outliers? I would say that about
(48:33) data collection and synthetic survey audiences. I would also say that about things like creative concepting, designing actual creative assets. The interesting things in those fields are not what looks most like the things that you can find that have gone before, but how do we break molds and patterns?
(48:56) But how about the marketers' perspective? Because if they're looking for user reactions, would they be able to get what to expect from the audience?
(48:56) Yes, it is by definition a backward-looking averaging machine. There are many good uses for it, and in most cases, I think it's probably perfectly fine to use it for
(49:28) synthetic audiences for survey work, for example. But again, most cases aren't the interesting ones. If I dig through fresh survey data, I'm actively looking for the pieces of data that stand out, not for the pieces of data that look like all the other data. And the piece of data that stands out, I then validate. I make sure that there isn't some mistake or
(49:56) other reason that it's standing out that it shouldn't, and make sure I asked the question properly and that the survey logic was working properly, and all of that. Once I know that this is a real result that stands out from the other results, then I start to look at what's happening and why. That's what's interesting. That's what turns into the reports that I write and
(50:15) the talks that I give. If you're just looking for the average of everything, if you're just looking for what the median person would say, and you create this audience that is a lot of people very close to that median person, it's a lot less interesting to me. I know that I'm oversimplifying enormously, and I know that there are potential ways of
(50:40) solving this problem. Personally, if they came to me and said, "We want to run your survey not with a real audience but with a synthetic audience," I would beg, borrow, and steal to avoid that happening right now.
**Concluding Remarks & Future Engagement**
(50:40) Well, Nate, we're going to have way more questions for you. Appreciate you coming
(50:58) on and answering quite a bunch, maybe some we didn't even know we had. As you have more to share, this is an open invite. I think it's a tremendous look that's not just, "Oh, here are some trends that are happening," but the stories behind them and what data you could trust, what you can't. It reminds me of some other areas I've
(51:23) seen research in, like research in millennials back in the day, where, "Okay, there are 80 million millennials." You can get the research to say whatever you want about them. They're both very into social causes and also very self-serving, just wanting to look out. You can find data to support anything with AI. See how it's so messy out there?
(51:45) Helping be a guide for what we can trust is eMarketer's heritage, it's your heritage. Glad there's a great fit together. Thank you for this, and I'm excited to keep getting to learn from you.
(51:58) Thanks so much. I'm happy to come back. As I said, when the data is ready, I will absolutely be posting it in your
(52:06) Slack and on various social channels, and I'm happy to come back and present it if that's of interest.
(52:06) Well, I think you'll get some takers here. No vetoes in this crowd, right? All right. Lots of applause going on. Thanks so much, and thanks everyone for always coming with your great questions and interest in the conversation today. I like how I looked
(52:24) away from the chat for a second, and it's like 20 new messages: "What the heck is going on here?" Nate, I could share that with you as well if you want to catch up on anything you missed. Thank you all. Next week we're off because I hope to see SEU and Architecture Live 3 in person, but then we'll be back with a great lineup of guest speakers coming up. See you all
(52:44) very soon.