The Battle for Verified Humans: AI's Data Quality Reckoning

Check out the full episode below! Enjoy The Exchange? Don't forget to tune in live Friday at 12 pm EST on the Greenbook LinkedIn and Youtube Channel!

 

The AI industry is rapidly consolidating around clearer business models and enterprise strategies. OpenAI is pulling back experimental products to focus on desktop apps and developer tools while exploring advertising and private equity financing. Meanwhile, Anthropic's device-control capabilities signal a shift toward hybrid AI that blends cloud intelligence with on-device execution.

Retail giants like Walmart are embedding proprietary chatbots directly into major platforms, revealing how brands plan to capture customers in an AI-mediated shopping future. Amid this race for market position, a parallel battle for verified human data is intensifying, with privacy-first measurement tools and behavioral intelligence platforms emerging as critical infrastructure for both research and commercial AI systems.

Many thanks to our producer, Karley Dartouzos. 

Use code EXCHANGE to get a 15% discount on your general admission IIEX tickets!

IIEX North America

IIEX Europe

IIEX.AI 

Stay Ahead of the Curve! Subscribe to The Exchange Newsletter on LinkedIn Today! 

Transcript

Lenny Murphy: And hello everybody. Happy Friday. So it's been interesting Friday morning here. Um, uh, Karen, unfortunately is ill. Uh, and we did not have time to pull in a, uh, a co-host. So I am flying solo. Although as always, Karley is in the background to try and keep it between the lines. Um, so I am, I really don't talk to myself that often. Uh, this is how this feels for me. So it seems a little. Awkward, uh, apologies. Um, but we wanted to make sure that we went ahead and, uh, got the news out because there's an awful lot going on. Um, so, but thank you for, uh, accepting the suboptimal approach of me just flying solo here before we, uh, jump into all the news. Uh, Karley, if you would, uh, let's get our plugs out of the way. It's how we pay the bills. So, uh, IIEX North America. Is right around the corner. It's going to be awesome. So if you have not registered yet, there's your codes. Go ahead and register. You don't want to miss it. It is the event. Then right behind that is IAX Europe. And there's your codes for that. It's awesome, too. Of course, they're awesome. And then already planned for the fall, we have our virtual event, IAX AI, which continues to be a great success, massive event. It is obviously changing as the world changes from an AI standpoint. So speaking of which, let's just jump in, because there is a lot of really big, interesting news that happened this week. And the biggest one probably was centered around the big platforms, particularly OpenAI had been kind of an experimental shop in many ways. They were shipping products over and over again. And one of the big ones was Sora, their video application. And I'm sure you guys remember when that first came out, and we're all watching it going, oh my god, that's AI? You can't tell the difference. Well, it's dead. They killed it. And the API is gone. There's no video in Chatsheep ET.

There is a billion-dollar deal with Disney that just is kiboshed. They're just done. What they are pivoting to, and this is indicative of where everything is going, is a desktop super app that unifies all of their different components, ShoutGPT, Codecs, to browser, around the Gentic autonomous computer work. We saw the same announcement this week from Anthropic that Cloud can now control your computer. So if we open that up a little bit more, that this focus now is from this cool cloud-based solution to something that is running on, still running cloud, but also running on device, and that's kind of the agentic component that came with CloudBot, et cetera, et cetera, and being far more practical oriented on utilizing these tools to do specific things and tasks for you, rather than the shift from kind of the original, Oh, look, we can write stuff and search stuff. And, you know, and all that's still there, but they're obviously moving towards being embedded architecture, embedded technology, uh, within individual users, technology ecosystems, either on device or within your, uh, your organization. Architecture for your cloud applications and being a personal assistant that runs everything Open the eyes cut their side projects. They are just focusing on coding and enterprises core business That's obviously because of the competition from anthropic They also hired a new ad sales chief from meta So they are obviously going to build a serious advertising business because that's another revenue stream on here There's an audience and there's eyeballs and that's a logical way to do it. When you're giving away the product for free, we become the product. Uh, so they're, they're doing that. Uh, the, uh, the gentleman from meta, uh, that's 200 billion ad machine. So open the eyes, obviously serious about doing the same thing. Uh, and of course, all of that ties in.

If you look at the, uh, the financials here that, you know, they're, uh, now shopping out private equity with a guaranteed 17.5% minimum return. Um, that's unheard of That's crazy. So rather than doing a traditional, uh, traditional raise while they're still prepping for an IPO, right, they are, uh, obviously leveraging from private equity with the enterprise application, because that's what private equity is going to be most interested in. Uh, so. To watch the macrodynamics of how AI is focusing on very specific capabilities overall. And on that news as well, Walmart abandoned OpenAI's instant checkout because of low adoption. Now they're using their Sparky chat bot and embedding that into chat GPT and Google Gemini. So Walmart's perspective, if you look at it, rather than letting the flow come to them via the AI search. They're actually embedding so they get to the flow. They are intercepting the potential shoppers and customers there via their Sparky application to build stickiness and adoption. That's a little inversion of the way that we've seen that. What's a little context on that, that Pulis' Commerce and eMarketer did some research that nearly half of consumers using AI while shopping would consider switching brands based on an AI recommendation. Independent assistants earn twice the trust of retailer-owned ones. So here's this play where Walmart is trying to take the position of their Sparky application will become the shopping aid, the, uh, the, the independent recommendation component within the Walmart ecosystem, rather than allowing, uh, Google or open AI to be that recommendation engine. So everybody's kind of angling to, to own eyeballs, own audience and own engagement, uh, at the device level. Now that brings up a whole other topic, uh, and that is Race for Quality Data.

The, at IIEX North America, Brandon Olish from Stanley Black & Decker, and Brandon, if you're out there listening, then shout out to you. We had, he's doing a presentation at IAX around data quality. He and I had a conversation yesterday. I won't go into the details on that conversation, but I will say that this week, the topic of Verified human data, verified human connection has been all over the place. And we see it in the news and I'm about to give you a couple of examples. We see it, I've had it in lots of conversations with different suppliers, with buyers. Brandon was one example of that. He is like the ninja master of data quality right now and really leading the charge I'm calling out the issues around data quality, and not just because of the impact on the individual project, but because of the impact that it has on the data ecosystem that brands are building. If you're building, let's say, synthetic, and it's off bad data, well, the synthetic is bad. And if you're loading it into your agentic system across your organization as part of a larger competitive intelligence platform or whatever the case may be. Uh, the contamination factor is just significant and severe and it cannot be ignored anymore. Brandon's an example on the buyer side that he is, uh, embracing that and is leading the charge. There's many other conversations that I'm having with folks as well on that. And that brings me to the other two, uh, uh, news points here that were interesting. Uh, remember. Or about two years ago, the launch of World. And that was this kind of strange sphere that existed in malls that scanned your eyeball. And that was backed by OpenAI. And it was a crypto component. But fundamentally, what they were doing was creating a database of humans. Now they released a really interesting paper that the need for private proof of humans, how verified human identity is critical infrastructure for democracy and human agency. And that's, that's interesting in an AI driven world that obviously has implications for research. I've often thought that, you know, world, uh, would be a great panel.

Um, because. Verified humans with an engagement strategy to allow people to participate in various activities is fundamentally what the the panel part of the research industry does and On top of that you go outside of the just answering questions and engagement But there's also the passive measurement component of that and that is growing in demand as well There's there's lots of buzz in the industry around some pretty significant projects that are related to passive data measurement. There's a lot of companies that are building capabilities around that to enable ethical, human-driven, identity-verified capabilities that's around, you know, take path to purchase, digital path to purchase, your shopping behaviors, et cetera, et cetera, and monetizing that directly. Of those companies, and actually I just spoke to two of them before this, DataSapien, which evolved from CitizenMe, shout out to Sinjin on that, goes back to this first piece we talked about on the shift to kind of device-based management of AI. DataSapien has developed an on-device AI behavioral intelligence platform for panels, but also for retail and brands and lots of different applications, where sensitive personal data never leaves the handset. It's privacy by design for the verified human data problem.

And it's a whole new spin. I'm calling that out because, one, I like Syngent and we've had this conversation. We've talked many times over the years back when I was trying to build Veriglyph with this similar concept. But the unlock there is that It is around creating technology that is privacy-centric, allows the consumer to control it, is virtuous, and gives them opportunities to monetize that in a variety of ways on their own terms, enabled via AI. And it's almost a sub-operating system for marketers or market researchers to engage with consumers in new ways. Now, think about that step back for a second. As an industry, one of our critical failings has always been, in my view, that we didn't think like marketers. We didn't think like how to create stickiness and engagement with consumers. Maybe community developers, you know, there's certainly that's not a blanket statement, but most of the time, we unfortunately have created consumer, have used consumers as a commodity and that doesn't fly anymore.

So I'm thrilled to see so many different technologies from outside the industry, like world, um, adjacent or inside the industry, like DataSapien, uh, or, you know, fresh right inside the industry companies like Curious and Behavix and Luth and M4 and Disco and, um, that, that are, Trying to build these solutions now that change that paradigm and and it's great to also see brands like Brandon from Stanley Black and Decker like Tia Maurer from from P&G and others that are also saying yeah, it's got to change There's too much at stake not just for the traditional research industry But now as we go forward data being the new oil will we need good quality oil? So interesting around all of that. Yeah, we'll move on on a couple other things, because I'm committed to keeping this short today, or at least shorter than it has been past couple months. Some interesting news, Kantar created the chief people and agent officer. So normally we don't talk about appointments so much, but this is an exception. That's the first major research firm to recognize that AI agents are organizational participants that require dedicated C-suite oversight. Uh, so hats off to Kantar to creating a new role for the chief people in agent officer, uh, treating agents as colleagues, as coworkers, as part of the, uh, I guess maybe not human capital productivity capital.

Lenny Murphy: Would that be the right way to say that?

Lenny Murphy: Uh, so that we're not speciesist, I guess. Uh, I think we're going to see a lot more of that. Uh, and it's interesting that the perspective is more around. Uh, working with. Working as colleagues versus simply as technology. So interesting perspective. So that was, that was cool. Um, the same time Kantar launched converser and AI moderated qualitative interviewing, uh, capturing experiences in the moment backed by 500 plus qual experts across 60 plus countries. So, uh, a hybrid, uh, overall. Oh, hello, Charlie Rader. Interesting that Kantar is launching this new solution that is a hybrid of human in the loop, qualitative overall, but AI-driven, and I think we're going to see more and more of that as well. All of the AI-moderated, qualitative, obviously, tons of success and adoption. I was actually looking at a new grid data that will be coming out soon. We saw a massive increase in adoption around the AI-moderated platforms. There is some shrinking that's happening on the human-led, human-driven, traditional qualitative. And here's Kantar trying to find a middle path. So hats off to them for doing that.

Lenny Murphy: Speaking of data again, Hanu from Behavix, which is one of the Behavioral measurement passive metering companies out.

Lenny Murphy: There was a sample con sample comes two weeks ago in in, Georgia Near my my old stomping grounds in between us in Atlanta And he wrote a really cool write-up on that I would encourage everybody to do it With the buzz, they're a sample con right sample con is an event for the sample industry all the providers all the suppliers They're all there. They're networking. They're cross-selling they're working with each other and and and trying to focus on those issues. And his argument, and I would agree with this 100%, that real human panels are becoming more valuable in age of AI. I agree 100% verified human audiences. That is the well, if we use the oil analogy. And that's, they're increasing in value. And they're going from survey distribution networks into behavioral intelligence platforms. And that is a, that was what I tried to do with Veriglyph back in the day, way too early, was to convince the panels that they needed to have a network supply of data, not just attitudinal, not just enabling attitudinal, but also behavioral. I'm glad to see that that is happening now. And that kind of ties into a neat pushback from Samuel Cohen at Fair on a Wall Street Journal article that said AI replaces researchers. And Samuel pushed back on that, that nope, we're actually seeing more research being done, not less. And even though some of that now is housed under synthetic. So maybe we look at the industry that's traditionally positioned and we think, oh, we're losing.

Lenny Murphy: Well, no, it's shifting.

Lenny Murphy: The spend is shifting. We always knew that was going to happen. As new technologies emerge and they change the way that we can do things, either through enhancing or through more efficiencies, then we see spend shift, always has been that way. Synthetic is an example of that. Samuel's in a good position because Fairgen does play in that arena. And the fundamental piece to reiterate this idea, this theme running throughout the show of the synthetic is only as good as the foundational data that goes into it, which means research panels, verified humans asking the right questions. Primary human data is the training, the core training component to be able to calibrate any AI system, but especially when we're dealing with any of the varieties of synthetic, whether it be digital twins or personas or, you know, or models or synthetic or the top ups, all of that starts with. Real humans with real data. Um, and it has to be quality and it has to be, uh, structured in such a way that we can unlock more value creation via AI in these new systems. And that's how we stay in the game as an industry. The spending will shift. It'll shift from a methodology standpoint, it will shift from a technology perspective, but the demand is only growing. So I'll get off my soapbox for that.

There are a couple other interesting things that happened from a launch. Cursory Technologies launched Athena, a unified AI control plane sitting between applications and AI models inside its RISC platform. That is an orchestration and governance play. And we've talked about that in the past few weeks, that fundamentally, a new way to kind of think about the industry, in all industries, I would argue, is architecture, orchestration, governance, and judgment. And the technology component, that infrastructure, that's technology. I would argue that data sits in there regardless of how it comes. The orchestration component, the ability to put all these pieces together. That's what we see the agentic revolution doing. They're orchestrators. This week, I have vibe coded six things this week. And what an amazing experience it has been. I'm blown away. And what's particularly interesting is in all of those, I would specify exactly what I'm looking for, and kind of put it in you know here's what I want to do etc etc and I see one agent it starts orchestrating the other agents and now we think about this that that is the definition of human in the loop I am defining what I want it to do what I want the outcome to be the agents are then orchestrating that for me that is I'm doing that in perplexity because it uses multiple models so it is an orchestration layer so challenge are made to kind of Think about that within your businesses. Where do you have the right to win? Is it in the architecture layer? If you're technology or data supplier, then I would say that's where you fit. The orchestration layer, that is certainly in governance.

Lenny Murphy: We'll kind of put those side by side.

Lenny Murphy: That is a human-driven but technology-led solution. That's not dissimilar to how we think about utilizing the right tools today. And then the judgment layer, which is fundamentally the kind of consulting component. And that is where our expertise, our IP, our imagination, our creativity, our intuition, all of those things play as well. And although each layer has value, the collective stack, I believe has the most value. And our industry, if you look at it, we're pretty much that collective stack in a lot of ways. So be thinking about that, guys. That's how we're going. Same variation on a theme, Informa TechTarget launched their AI Visibility Audit and GEO Topic Planner, which is trying to play within that infrastructure component, so they're part of that tech stack within the infrastructure layer, and help brands manage discoverability in AI-mediated environments. They're orchestrating solutions to fit within the marketing application layer within that tech stack. Back to Virtual Personas, Savanta launched their Virtual Personas, AI-powered synthetic personas integrating the Ocean Big Five personality traits in Schwartz theory of basic human values. Big fan of that as well, so it's not just the data, it is also understanding those implicit components of personality value drivers. I've seen some other solutions out there. Our friends at Sparky AI, they've built that into their marketing platform as part of their synthetic, their personas. So we're trying to build this more holistic modeling model, if you would, on personas and synthetic that's not just who, what, when, where, and how, but also getting to the why and understanding those drivers.

So lots That's all the news pieces. There are a couple things that I would suggest that you guys read. There's a great article on the specification economy. And this argues that the most valuable skill in every business function just changed to the ability to specify precisely what you want, what good looks like, and evaluate AI outputs. I can testify from my own experience. Week now of you know, finally finally diving into the the vibe coding building piece 100% 100% builds amazing stuff really fast, but it's not always right. It's not always good. And of course, I have to give a direction so As we I would argue that in our the world that we are in now saying the emergent world it is the world Researchers and strategists Think about that. Think about how we, that is that judgment and some level of that orchestration from kind of a human perspective, how we translate that into impact, utilizing these tools to do the heavy lift. So definitely worthwhile. Also a great example of something that everything old is new again. Chameth, I'm sure that you guys all know Chamath, Silicon Valley investor, et cetera, et cetera, podcast host. Dived into prediction markets. Now, I remember working with prediction markets many years ago, my friends at ConsensusPoint, shout out to Brad Marsh, and others that were building prediction markets, even System One, when they were brain juicer, used prediction markets.

Lenny Murphy: And you couldn't give it away. I mean, it was a hard sell.

Lenny Murphy: Even though arguably the results were more predictive of outcomes. Um, and the whole wisdom of the crowd, you know, aspect of that. And boy, it just exploded and exploded through the betting markets. So now, uh, $6 billion in weekly trading volume, 100 growth in 24 months through The big prediction warrants. It was a Kashi and I forget all the names anyway Now sure I was Betty But not all of it that high accuracy in predicting the outcomes of elections The data feeds from those are becoming valuable and inputs into from, you know, traders, et cetera, et cetera, to understand what's happening and predict outcomes. There is a magic that happens there. And hey, James, glad you agree. And yeah, Nancy, the brain juicer offices were in Manhattan. I remember going there with Susan Griffin at that point when she was still the CMO of Brain Juicer, fun times back then. But it's something to watch because the, we had an opportunities in the industry to really get into that and drive it. And we didn't. Now it is a multi multi-billion dollar sub industry that many people are utilizing to replace some traditional research. Uh, and as we look at this idea of AI synthesizing information, building synthetic personas.

Lenny Murphy: It's all around simulation, foresight, decision market capabilities.

Lenny Murphy: All of those things are synthesized through these data feeds that now the prediction markets have hit scale.

Lenny Murphy: They're not perfect.

Lenny Murphy: There's certainly a chicanery that happens in them.

Lenny Murphy: The models are pretty good.

Lenny Murphy: The track records are pretty good.

Lenny Murphy: We could have owned that as an industry.

Lenny Murphy: We didn't.

Lenny Murphy: That ship sailed.

Lenny Murphy: But I would definitely encourage, as we are thinking in this world of orchestrating and pulling information together to address business issues and using our own innate skills as researchers, as DOC connectors, we give it another look and think about where the prediction markets fit into our toolkit as well.

Lenny Murphy: There's still platforms out there that enable prediction markets that function more like kind of in the survey approach to things.

Lenny Murphy: And there's also the big platforms out there that can be used.

Lenny Murphy: So keep your eye on that.

Lenny Murphy: Follow the money.

Lenny Murphy: We always say a hell of a lot of money flowing into those platforms.

Lenny Murphy: Both from investors and obviously in the daily trades.

Lenny Murphy: It is engaging, it's interesting, it's fun.

Lenny Murphy: I set aside any moral arguments about betting.

Lenny Murphy: And I know some people do have that.

Lenny Murphy: I do as well.

Lenny Murphy: I do not enjoy gambling because I really don't enjoy losing.

Lenny Murphy: And so I don't do that in that way.

Lenny Murphy: But I totally get it.

Lenny Murphy: I get the fun and the engagement and unlocking maybe some type of innate human intuition, the wisdom of the crowd, seeing how other folks are doing things and how that can be a highly, highly useful and predictive data feed as we think about trying to address business issues of understanding the world today and understand where it's going.

Lenny Murphy: and that's a core part of the research industries function.

Lenny Murphy: So that's it. I've rambled on enough. Thank you guys for putting up with flying solo. I hope that you have found this useful. Thank you to to Charlie and James and Nancy for hanging in there and sending comments so I know I'm not here just talking to myself. Thank You Karley for keeping it between the lines. Join me please and wish that Karen gets well soon.

Lenny Murphy: She she you know was at Qualtrics last week and you know travel etc, etc I have a lot going on and this week it just hit her and We want her to be on the mend and be back next week So we can actually have a real exchange not just a soliloquy But that's it for today. Have a wonderful weekend and We will talk again very soon. Thanks a lot.

Links from the episode:

OpenAI exits video entirely 

OpenAI cuts "side projects" 

OpenAI hires a new ad sales chief from Meta 

OpenAI offers private equity a guaranteed 17.5% minimum return 

Anthropic's Claude can now control your computer 

Walmart abandons OpenAI's Instant Checkout 

Ipsos Data Drops: How Americans feel about AI shopping agents 

World.org publishes "Private Proof of Human" 

DataSapien launches an on-device AI behavioral intelligence platform 

 Kantar creates "Chief People and Agent Officer" 

Kantar launches Converser 

SampleCon 2026 recap from Behavix  

Samuel Cohen (Fairgen) directly rebuts the WSJ "AI replaces researchers" narrative 

Cursory Technologies launches Ethana 

Informa TechTarget launches AI Visibility Audit and GEO Topic Planner 

Savanta launches Virtual Personas 

"The Specification Economy" 

SampleCon 2026 Long Read Recap from Behavix 

Chamath Palihapitiya Dive on Prediction Markets