AI vs Human Research: The Shocking Results That Changed Everything

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!

 

Traditional market research is being challenged by synthetic alternatives that offer comparable accuracy at significantly lower costs. This shift raises fundamental questions about the nature of authentic insights and whether the source of data matters if the outcomes are reliable.

As brands increasingly demand cost-effective solutions with faster turnaround times, the industry must grapple with balancing efficiency gains against potential loss of human nuance and contextual understanding.

Many thanks to our producer, Karley Dartouzos. 

Use code EXCHANGE30 to get a 30% discount on your general admission IIEX tickets!

IIEX.AI

IIEX APAC

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

Transcript

Lenny Murphy: All right. Hello everybody. Uh, welcome to this week's exchange. So, uh, Karen is on a much deserved vacation in beautiful Brazil. Uh, and her husband, Tim, who we advertised was going to be stepping in, unfortunately is so busy with his job of change management that he got called into client calls. So, uh, you're stuck just with me. We'll see how this goes. It's not optimal. There's not much of an exchange other than me exchanging with you, but hopefully we'll cover some news and I'll obliviate on a couple topics and we'll call it a day. But do let us know what you think, whether we get caught in a bind like this again, whether a solo, flying solo is good, or if you're like, no, never do this again, Lenny. We actually only enjoy this when you're talking to somebody else versus just you know freewheeling by yourself so on that note as always Karley is in the background and she will be posting links as we go and let's uh let's run through some news and then we'll tie some threads together here so first insights analytics industry news congrats to dig insights they acquired one click to deepening their cultural intelligence and AI capabilities. They continue to grow.

That is interesting as they're expanding capabilities, particularly this idea of cultural intelligence that is driven by AI. I think that a big piece of this AI race right now is finding a differentiator with different datasets effectively that have a different angle. In this case, it's looking at the kind of cultural nuances and how that informs decision-making from an Insight standpoint around messaging products, etc, etc. So, congrats to them. Also this week, Insight Association reported their annual survey, Strong Growth for the Fast-Changing U.S. Insight Sector is the headline. The report shows 7.6% U.S. Industry growth to 82% 2.9 billion shifting to tech and digital. Is there any surprise with any of that? I doubt it. It's certainly indicative of the shift. Now, one thing we should point out is that the IA report is actually a year back. So what the recording is effectively, you know, previous results. So anything that they're reporting now, know that that trend was happening a year ago, and it theoretically is continuing to accelerate, and we'll know that as they measure that later on.

So, but overall, good news for the expanded view of the insights industry. As always, there are some surprising companies that are included in that, particularly the big consulting companies, Accenture, McKinsey, etc. They are part of the industry now, and I think they're part of the dynamic of how things are changing. But we'll actually get to that a little bit more in a minute. There were some cool product launches that we saw. Data Quality Co-op has continued to enhance their platform to strengthen insights, transparency, and access. So they are now using historical metrics, supplier scorecards, custom groups, and granular reporting. Of course, if you're not familiar, Data Quality Co-op is focused on data quality and a new model of consolidating and curating information across the industry and your particular supplier set in managing data quality. Check that out. Related topic, Dynata expanded their partnership with Data Axle to advance identity resolution data enrichment and media activation. So that's an example of one of these trends we've been talking about for a while of panel companies that sit on an awful lot of profile information, and increasingly behavioral data as well.

Now, going out and connecting to other data sources and looking at getting to first-party individual-level data that expands the overall view of the consumer, that's something certainly I had played with for a while in Verigliff. I know that it's hard to mirror those data sources, it's hard to do identity resolution, it's hard to do that in a first-party permissioned ethical way. Hats off to Dynative for continuing to do that, to look at ways to expand the utility and view of their data. This wasn't mentioned in the press release, but my bet is that the angle there is moving towards kind of an always-on data asset that will be feeding into various and sundry AI solutions, as well as increase in quality and targeting in kind of the current more transactional model of research. So those things kind of jumped out. A few overall tech things. The startup Lovable, the Vibe coding platform, just raised $200 million dollars at 1.8 billion valuation.

They're a Swedish startup, although a startup with 1.8 billion valuation. They really have tried to scale that idea of vibe coding. And if you're not familiar with that term, basically you describe the objective of what you're trying to achieve from a technology standpoint into a specific application. And AI basically writes the code and creates the entire platform. I actually was talking to a good friend of mine earlier this week who was helping his daughter launch a company. And the app itself is they were building that using Lovable. And then specifically within the app, there's a gamification component. And the daughter, who has no experience with coding in any way, shape, or form, created, I forget the number, but tons of games to be embedded into the platform over a weekend just using this Vibe coding platform. I think the important thing to recognize here is we've talked about the barrier to entry for technology development is decreasing. These Vibe coding applications like Lovable are an example. So if you're thinking, oh my God, how do I invest to develop some new technology or new solution internally? It's going to cost, you know, a million dollars in six months. That is not the world that we are in. The world that we are in is through these vibe coding platforms.

You can do an awful lot of the heavy lift. I don't know if it's 100%, but certainly a lot of the foundational work literally within a few hours. And that also decreases the barrier to entry for new companies entering into the market. That's why we are going to see an accelerating rate of new solutions that have variations of different pieces of the process of insights coming to the fore, some from existing suppliers, from new folks coming in, because it's just a heck of a lot easier to develop something and get it into market now utilizing platforms like this. So pay attention. On a related note around data, two senators, Hawley and Blumenthal, have introduced an AI protection bill. Now, as we record this, there were a lot of announcements yesterday around executive orders from the Trump administration regarding prioritizing AI development and scaling. Some aspects of that seem to skirt over issues around data ownership and privacy, which are concerning.

So it's good to know that there is something that is happening legislatively, whether it passes or not, we'll have to see. But people are paying attention to the idea that in the era of AI, your data, your likeness, your feeds have some level of IP and are protected, or should be protected and crafting some legislation to ensure that it's not just the free-for-all of capturing training data from anything and everywhere and cloning that and creating new things off of existing solutions or existing owned assets. Pay attention to that. I suspect that's something that the Insights Association will get behind as well in trying to support that. It goes along with Obviously, our overall focus as an industry around protecting the rights of individual respondents. In this case, it's writ large in a whole other way, but it will be part of the world that we live in and thinking about within the insight space specifically, how data is protected and people are rewarded for sharing their data in order to drive site solutions. McKinsey had a really interesting article on seizing the agentic AI advantage. And the point here is going from AI pilots to scaling proactive AI agents for business processes. And to follow on, there was another article. This was from just a pundit on AI market clarity, on kind of really where the things stand right now with AI markets. Who are the leaders? What are the future disruptions and sectors? And I wanna package both of these up.

I encourage you to look at both of these because I've been doing an awful lot of thinking. We talked about it last week with Karen. We're shifting away now from the idea of just the research process being disrupted by AI, and maybe it's augmented, maybe it's disrupted, maybe it's recreated, there's lots of ways to think about that, but the research function sits within a business, and the business process itself is what is beginning to change, and that is why I'd hoped that Tim Lynch would be able to join us this week, because he works in change management around technology, and definitely has a point of view on what this pace of change and adoption looks like, The sheer fact that he couldn't be here was because he got pulled into client meetings on these issues. And I've heard anecdotally that he is incredibly busy right now. So the pace of this change of the fundamental business process is accelerating. Over the past week or so, I've had multiple conversations with senior leaders at many companies, uh, which is kind of how I spend a lot of my time, uh, who have confirmed that this acceleration from clients is happening.

Um, clients are now pushing their suppliers across the entire organizational structure, not just insights everywhere. To now integrate kind of the primary business processes into an agentic infrastructure that they are rolling out. So think about that from a procurement standpoint. Think about the executional elements of that. It's really profound. And there's actually a blog post that will be coming out in the next week or so by me. Where I really played with that idea and used AI to kind of think through what the day in the life of a client-side researcher looks like in this rapidly approaching agentic future. The point is this McKinsey article and the article on the market clarity are really strongly indicative of the view of where things are now and where they are rapidly going. Some of the folks that I've spoken to that are in the middle of trying to catch up with this, they're talking about horizons of literally a few months to be able to deploy these solutions and how they are scrambling internally to transform. I had another conversation earlier today with a long-term friend who's involved more on the capital markets side of the industry and has a long history of working on deals on that front. And what he is seeing as well as following the money is this push for, there's time for existing suppliers to adapt to this, but not a lot because there are new companies emerging that are much more linked to this kind of a genetic process and they're getting a lot of attention right now.

So if I had to guess Here's how this I think this is going to play out I don't think there's gonna be one ring to rule them all I think they're gonna be some companies on the buyer side that haven't already invested very heavily into a into a operating system across organization based on agentic AI, a lot of that is going to be Microsoft driven, Google driven, maybe Amazon through Amazon Web Services, you know, large infrastructure players that are providing all the pipes, right? The foundational elements necessary to transform business processes. That's probably where this is ultimately going to lie is integration with those platforms and how they are used to enable the entire organization and the insights organization to centralize the operations of their particular areas of interest through a series of executional agents that are built into what will basically be the browser. And we talked about that last week. There's another layer that I think is likely, and that is big enablement platforms like Salesforce, for instance, or maybe it's HubSpot. Those kinds of standard leaders in specific areas of business that also have an agentic ecosystem. So they're plugged in with a series of basically a marketplace of suppliers that are aligned to their area of interest. So let's, if we use the Salesforce example, the CX platforms, that would be logical, that already exists, but for that to be agentic, so you would log into your Salesforce and Oh, I want to understand my customers from the existing data that's available to them. And you go through some hypothesis testing based on the existing data.

Lenny Murphy: That is basically synthetic data.

Lenny Murphy: And then through there, OK, well, now I want to go through and actually do a survey to my customers to understand this with real live information, and all that is executed through a series of prompts through the agent. Behind the scenes, excuse me, everything is executed without engaging with people. The AI in that scenario would make the recommendations for the survey, write the survey, then it would connect to whatever the CX platform is that you're using, let's call it Qualtrics, just for example, but it could be any of them, then that the Salesforce agents interact with the Qualtrics agent, it's executing the survey to your customer list, all the data's coming back directly into your Salesforce dashboard, and that is feeding into, maybe a connection there, into your broader Azure platform. That's the world that we are talking about.

Another scenario is a trusted partner. And that's probably the most viable and interesting part for our industry. If you are, you will either be a part of an ecosystem and or actually it's probably a better way to say it than either. You will be a hub for an ecosystem of various solutions that are driven agentically. So it's thought of as kind of a tiered platform, a series of agents that are interacting with each other that are executing a ton of research without direct human intervention in many cases. Now there will certainly still be opportunities where there is human intervention where it needs some more some more thought being put into it in terms of, you know, design and analysis, especially as we get into recommendations for action, kind of the, you know, the so what now what type piece of things, but that's going to be a thin sliver of the business.

The bulk of the process, which defines a lot of the business models for research companies are going to simply be executed via the syngenetic layer, and that is happening very, very, very quickly, that transformation, because it is easier for organizations, B2B organizations, to validate that. The cost savings alone to be able to take out a lot of the currently necessary processes on transactions that don't really deliver value, they're just something you have to do to get to the value, which is going to be the answer, is that it is streamlined overall. Brands will save millions and millions of dollars in human resource costs. So folks are actually focusing more on rather than managing the research process to manage the research outcome. So with the type of numbers and efficiencies that are being talked about, particularly in the McKinsey report, there's no way around it. The ship has already sailed. And I think that's the takeaway that I hope that our audience gets is, as we're looking at the industry as a whole and looking at these trends, we're trying to communicate them to you.

So as leaders, you can start adapting your business for this transformation that is happening now. It is not a future state. It is happening right now. And when we, the new stories that we pick, although we don't necessarily pick up as far as, we try and get as much in here as we can. We don't leave much on the table, but increasingly they paint this picture. And that's why we do this every week to give you the understanding of all of these variables that are and all these different dimensions that are coming out to give you a sense of exactly what's happening and where that's going. So anyway, those are all some important things, and Karley has put those links in. There's some recommended reading, and then I'll wrap this up. There is a great article from the Harvard Business Review on using Gen I for early stage market research. This is effectively talking about synthetic samples.

People have shared with me examples of their synthetic sample experiments that, compared to live respondents and or to existing benchmark data, are now achieving well over 90% accuracy. So that is dependent on using real customer data. So keep that in mind. But the need to ask the question of a real person is, for some things, decreasing. The synthetic sample, when it is quality based on quality inputs, is increasingly able to answer many of the questions that currently we use for custom research. And that's just where it is. Whether we like it or not is kind of irrelevant. Last week we posted a link to Mark Ryan's article on this and I think he's exactly right. We simply have to adapt. One of the leaders I spoke to this week brought up that you know how to price for that versus a more transactional process we currently use. Got to get ahead of that guy. You know this looks this kind of always on a genetic approach probably looks a whole lot more like a subscription than it does Anything else and the price points are different.

Those things can play havoc with businesses If I had to hazard a guess many folks in the industry particularly on the service side You're going to see a decline in revenue, but an increase in profitability. And those that make it through that process will see potentially an increase in volume, which will get back to revenue growth in terms of volume. It's just a bigger kind of constant flow of money flowing in through basically And a lot of research is going to look like that overall.

Lenny Murphy: So there we go.

Lenny Murphy: There's another piece here that really spoke to me. Phobia is this fear that equals cheating. I've been very open in talking about that. I think I've gotten over it to a great extent. But there's still some things that I feel like I'm cheating if I use AI. So it's a very provocative piece that challenges that perception, that it's really not cheating, because in that world, we are focusing on process, not outcome. And if the outcome is where the value really is created, as long as the outcome is high quality and good, then it's not cheating to create efficiencies in the process. Interesting things that we're grappling with. There's also an interesting article in the New York Times on AI versus AI. This is really looking at data quality in a very broad sense. In this case, looking at marketing and how AI is being used to try and catch cheaters, but that applies to our industry as well. We've talked about that before, companies that are focused on data quality are deploying AI tools to help defend against bots overall. So a really interesting article in the New York Times.

On a related note, a social trait study revealed that Americans overestimate their ability to detect AI-generated misinformation. So as we think about how we can spot these things, we can spot these AI fakes. That is getting harder and harder and we may be overestimating our ability to detect what is AI, particularly what is AIBS versus what is actually true. We probably need to continue to invest in utilizing these AI tools to help detect AI fraud because it's really hard. It's really hard and it's getting harder to separate the wheat from the chaff. And then finally this was particularly interesting on chat GPT psychosis and LLM psychopathy There's been a lot in the media recently of that most of the models are trained to give you the answer that it thinks you're looking for and also to be engaging so that's just economics wanting people to use it. In some cases, maybe they've gone overboard. And there are examples of some users that are experiencing some mental health issues as a result of that, that not only the folks that think that they're in a relationship with their particular iteration of AI, but also reports of the AI maybe feeding into some of their own delusions a little bit more, in some cases, maybe even helping to create that.

Lenny Murphy: So that's, it's an important component to recognize that maybe we go a little too overboard in some of these solutions.

Lenny Murphy: Again, as we record this just yesterday, there were several executive orders that were put out.

Lenny Murphy: Some of them had to do with AI training to try and make them maybe more objective, a little more business-oriented, versus the current models that have been maybe a little more focused on a kind of touchy-feely and engagement.

Lenny Murphy: We'll see how all that plays out.

Lenny Murphy: But it is a phenomenon.

Lenny Murphy: Which, just to wrap that all up, guys, we continue to just be in a brave new world.

Lenny Murphy: There are so many things happening in so many different angles and dimensions around these technologies and how they impact every aspect of our lives, but particularly from a business standpoint, and particularly as it relates to insights and analytics, because we are a data business.

Lenny Murphy: It's important to stay abreast, pay attention to these things, because the world's moving really fast.

Lenny Murphy: I think there's a Ferris Bueller's Day off quote in there somewhere, but I can't quite pull it out.

Lenny Murphy: So on that note, we will be back live next Friday.

Lenny Murphy: Currently Susan Griffin is gonna be joining me because Karen will still be on vacation.

Lenny Murphy: That's the plan.

Lenny Murphy: Hopefully you're not stuck with me again During this solo because I don't particularly think this is that much fun But let me know what you think in the comments, please as always if you're watching this on on YouTube like and subscribe if it's on LinkedIn, make sure these sign up for our newsletter and That's it for this edition of the exchange.

Lenny Murphy: Everybody.

Lenny Murphy:Take care.

Lenny Murphy: We'll talk to you next week.

Lenny Murphy: Bye-bye.