Market Logic & IPSOS: Qual360 Keynote Presentation

We wanna share a little bit how we see AI shaping the future of market research and insights. I’m super happy to be on stage together with Aynsley, from from Ipsos. Aynsley will share, insights from Ipsos themselves about how this, happens, in the market, but also how Ipsos themselves are using, AI for research. And I will add to that then a little bit what we see as a vendor, of course, within our customer base, what are problems, and where can AI help to drive things forward, not replacing people, spoiler alert, but rather empowering them. Before I hand over to, Aynsley, a very brief, intro to Market Logic, what we do. I’m sure Ipsos does need introduction. We are a vendor of software platforms. Our customers use our software to onboard all the research they have to to those platforms, be it their in house research, be it syndicated or other research, and then they use that to, bring it to their stakeholders to help them answer questions, with the existing knowledge estate that they have and to drive insights and and bring them to life and decisions in the business. So that’s what we do. But now without further ado, I’ll hand it over to Aynsley to speak about Ipsos’ learnings and views. Thank you very much, Olaf. So, yeah, my name is Aynsley Taylor. I’m the director of the Ipsos Knowledge Center. I imagine most of you know, or at least have heard of Ipsos. So hopefully that we don’t need too much of an introduction. But, we’ve been working with Market Logic for nearly a decade now, I think, on various projects in that time. At the Ipsos Knowledge Center oops. Sorry. I better go to my slide, hadn’t I? Silly me. The Ipsos Knowledge Center is a team of about, which I run, is a team of about fifteen people in five different countries around the world. I want to make a disclaimer actually because I’m stood stood in front of a room of experts, experienced researchers here. I am not a quali. I’m probably, arguably, not even a researcher. And I’m certainly not an expert in artificial intelligence. So you’re probably wondering how I managed to, block a professionals here. I mean, the the truth is that my job puts me at the intersection of all three of those things. So, hopefully, I will have some useful observations to offer you today. And thank you, Olaf, for, giving me this opportunity to speak today. I appreciate it. So, generative AI. We’ve all heard of it. Right? So you might be sick of hearing it, a bit. But, I mean, for the last two years, nearly two years now, everybody’s gone a bit crazy on this topic. The promise, the potential, the hype, the scale of the investments in it has been off the charts, quite frankly. And some of the biggest companies in the world, you can see mentioned here, at least by market cap, if not by revenue, have really, really gone to town on this subject. Because after the, initial excitement sort of tied down a little bit, we started to think about some of the risks, some of the potential limitations possibly of the technology, the bias that’s inherent in some some of it, and the hallucinations that sometimes, we see, and the security risks and what happens to your data when you put it into these large language models. So so far, so what? You’ve heard all of this before. Right? But I just wanna set the scene a little bit more of scene setting here. So we, last year, every year in fact, we do this AI monitor every year since Cross AI became a topic of discussion. So we run this survey in thirty two different countries all around the world. It’s online, of course. And the most recent one was last summer. So this is a you know, it’s a fast changing topic. So you might want to take this with a little pinch of salt because six months on attitudes might have shifted a little bit. Although, we we haven’t really noticed them change too much, between twenty three and twenty four. But we’ll find out more this summer about where people are, then. And so just a few highlights from this. At a glance, what did we find I mean, these people are your respondents. Right? So this is more than just general context, and, hopefully, it will be interesting to you just for that reason. But these are the people you are researching what they think about the technology. So two thirds of them around the world, on average, claim to understand what AI is, but only half say they understand which, professionals services typically use it, which seems plausible to me at least. Generally, in most of the countries that we asked, people tended to thought that humans were more likely to discriminate than, artificial intelligence and large language models, which interesting finding. Two fifths of us don’t think that companies can be trusted with our personal data. And, nearly half of Gen zed I know we’re all a bit obsessed with Gen zed, but, they’re as cheery as ever these people. They expect AI to replace their jobs fairly soon. Although, this proportion decreases as you go through up through the generations to older people, which, you might expect because possibly because of the sorts of jobs they do and perhaps they’ve got a little bit less career in front of them to worry about. But, of course, the devil is always in the detail with with these things. So if we look at the difference between regions and countries, I know, most of you here will be working across Europe, but some of you will be working, beyond that in different parts of the world as well. You can see huge amounts of diversity and opinion on some of these questions. So for example, on people’s awareness of the products and services that have AI bundled into them. You see Chinese right at the very top there. They you know, most Chinese people say they understand what, products and services have AI inherent in them. But right down there’s a big big drop off right down to the bottom to Canada, where the Canadians are about thirty six percent. Will say the same. And if we look at Europe specifically, you’ll see that, of course, Europeans tend to be a little bit more, measured, skeptical perhaps, or claim or or certainly more cautious, I would say, on this question. So the the European countries in red and the the others at the bottom are, North America, New Zealand, Australia, Japan. So more developed markets tend to claim less awareness of these things. It’s a similar story, with the assessment of whether there are more benefits than drawbacks. So, again, we see European countries typically, a little bit more skeptical, a little bit more wary. So anywhere less than a half, generally, of Europeans will say they, think that there are more benefits than drawbacks, whereas people in the rest of the world tend to be more enthusiastic. Excuse me. So just to dive dive into a little bit more of the detail of this, this study. So if we if we take a close look at the security point in particular, opinion is actually pretty split overall. And about half of the countries go one way on this question and half go do this. The jury’s really out on this around the world about whether, people think their personal data is safe in AI or not. Younger generations, again, probably won’t surprise you to hear this in spite of there’s a bit of cognitive dissonance here because although they’re a little bit worried about their jobs and careers, they’re actually more likely to trust the technology, which I guess might be because younger people, perhaps they’re braver, more adaptable, or perhaps just simply more familiar with technology in general. Who knows? Some qual might help actually in in unpacking this particular finding. Half of us say that it’s already changed our lives, which maybe that’s human beings showing a lot of self awareness there, probably more self awareness than the machines do, at least for the time being, we hope. And two thirds of us, are expecting AI to impact our future, which, I think is probably an under underestimate, or at least I would say. I’m surprised that’s not higher. So it’s a fairly even split between those who say, the product services make them more nervous or those who make them more excited. So, yet again, the jury’s out on this question. Obviously, these numbers add up to more than a hundred percent. That’s because it is possible to be, nervous and excited at the same time as, of course, I am at the moment living proof of that. And, if we look at some of the cultural differences on this question, again, around the world, forgive forgive me. I get quite excited about doing charts like this. But, along the x axis, you can see people getting excited. And on the y axis, you can see people being nervous. So, what we find here is in Asia, people are moderately nervous about this, but actually really quite excited about the, prospects for this technology. But if we look, on the top left, people who speak English as their first language like me and perhaps like some of the others in this room, there’s something about speaking English which makes us more nervous, tending to be more skeptical and wary and less excited. And and in Europe, people are much more measured and sensible, obviously. And I think that’s probably the best place to be. Yes. And on the job question, just returning to that one briefly. So just to add some nuance to this particular point, people, who are more educated actually, also, are more concerned about the future and they suspect more concerned about the future of their careers. Perhaps more knowledge makes you more anxious and and as you are professional researchers in this room with a lot of knowledge about a lot of things, maybe you might agree. I don’t know. And given that the aims one of the aims of AI is purportedly to liberate us from drudgery, maybe there’s something counterintuitive in this finding as well. So anyway, all well and good. I’m happy to talk about more detail on these findings. It’s all it’s all on on the Internet as well. It’s all up there on our website. So, if you want to explore it more, you can you can look at the report. I’m happy to send it to you. What does Ipsos itself make of all of this? I mean, what type of position philosophy on this? So it’s very similar to Market Logic, actually. We believe in the blend of human intelligence and artificial intelligence. AI, they are they are cliche. AI will not replace your job, but somebody using AI might do. So, I mean, the one of the key points, I think, in insights this manifesto is actually about data because what Ipsos, I think, should claim and sometimes does claim is that we have lots of data. We’re really good at collecting data. We’ve got lots of it. And, actually, any large language model is only really as good as the data you put into it. Right? And that’s one of the, you know, one of the reasons we see lots of bias in these models because, actually, the data itself might be biased in certain ways. So we we think we’ve got the best data. Therefore, we can produce the best, the best AI powered service as well. And just to cover some of the thought leadership we’ve got on this, again, it’s all been published. It’s out there. You’re very hap very happy to to share it with you if you’re interested. This is all I mean, we’ve got tons of this stuff, but this this is actually by our call teams. This is the Ipsos EU service line who’ve written all of this. So we we started eighteen months ago. This series, Conversations with AI. The first one, Iterative Sciences, about prompt engineering. Things obviously moved on quickly since then, but I think we’ll stand by this one. So the the first one, the prompt engineer. The second one, we talk about the strengths and languages of excuse me. The strengths and weaknesses of, for transcription and translation, how it works best with different languages, and we, of course, emphasize the importance of having a human in the loop, which actually comes through very strongly in all of these papers. Its power, you know, its its its utility for ideation. Machines do think differently to human beings. That can be a good thing sometimes, provided we understand, exactly the limitations of that. And they can sometimes help to unblock the creative process or unlock some left field ideas. This one’s more in my personal ballpark. It’s about curation and knowledge management and using it for synthesizing, structuring, indexing, and summarizing information. But, again, emphasizing the point that humans are needed to build some of the connections, find the new ones, and tell the stories around the the findings. Personas were mentioned in the in the, in the last session. So this one speaks to AI persona bots. And, again, they do have which are trained on real customer data, and these persona bots do have some real, strengths, but they lack empathy and emotional understanding, which, of course, is absolutely integral to whether you guys do. So, you know, they they do have use, but, it’s all of this is about understanding exactly how they can help you and where you need to be careful. And lastly, I think we’ve we probably grew more skeptical as this series went along. This this last one about using, AI for moderating. And, again, the the we this one’s probably, I think, the most, the least enthusiastically six paper because we find that, you know, the moderating at the moment, at least, it’s not really something that, that we can use AI for because it struggles with improvisation, often loses focuses on research objectives as well. Anyway, so all that said, here’s some of the things we’re actually using this for. Our our core teams are using this for around the world. We’re using it for curation, workshops, moderation, and storytelling. I say using it. I mean, it’s probably still an exploratory phase at the moment, but, it it is something we’re investing quite a lot of time and energy into. Quick word on the Knowledge Center, which is where Olaf and I really, really do come together. So this is the stuff that I manage personally, and this is our, this is our site that we have internally, which contains all of thought leadership, all of the material that we built over the years to help our researchers and all the different teams and specialisms, qual, quant, and everything else, to sharpen their professional skills and to prepare for client meetings. And as you can see, the little blue bit at the top there is DeepSights. That’s the, the AI engine that, the module that our last team has built. Excuse me. Now this not everybody in in the, in the organization has access to this yet, but the people who do have it have been very, very impressed. I’m not saying it’s just because our last invite him to go on stage, but I promise you this is true. It has been very, very useful. And, one of the reasons I think just to, just to pat myself on the back is actually because the governance of what goes into this is really, really strong. We have a team that that actually manages what goes in, and that includes the application of metadata and making sure that things are structured in the right way, written in the right way, and, actually setting standards on what goes in. So, it’s not rubbish in, rubbish out. It’s quite the opposite. It’s it’s high quality in, high quality out. And, I think some of the, some of the evidence of this can be seen in, something else we do call it, Sosfacto. I don’t know if any of you guys have heard of that, but that’s our in house, bundle of large language models that we’re using. It’s in more of a general purpose tool. And that’s at the moment, I think that the contrast between Ipsos and Ipsospectral and DeepSights is, really quite interesting because DeepSights has very clear application. We’re using it for one purpose in particular in the organization. That really, I think, has helped it, achieve a great deal of impact. Whereas Ipsos Patch at the moment is a I would say, and my boss would not thank me for saying this possibly, but I think it’s a solution looking for a problem at the moment. And I think that one of the lessons that we’ve learned through through, experimenting with this technology over the last year is that, actually, you do need to have a very clear use case for using AI. It’s not something you can just spray around, use willy nilly, and hope, it will it will help you because it won’t. And I think that’s just about all I’ve got for now, Olaf. Back to you. Yes. Thank you, Aynsley, for those findings and observations, and that’s, I believe, very interesting. And let me know take it one step further or add to that by talking about what we see from our perspective, how our customers, you know, what challenges they face and how they can use AI similar to what Ipsos is doing to to get better and to bring insights more to life. And I wanna start by a little scenario. Imagine a while ago, somebody came. I had the sentence, I, need a survey, and you went out and did the survey and came back with all these beautiful powerful insights, deep understanding of the consumers, a lot of further understanding maybe of evolving attitudes, a lot of nuance. That’s all available there. And for sure, that will help inform your stakeholder who you were working with on the project. But also, there’s a lot of insight that would be so valuable to so many other people in the organization. And imagine at the same time, somewhere else in corporate, product innovation team is thinking about their pipeline or maybe the marketing team is thinking about campaigns they might wanna do. And their additional information, additional lens on on the human element would be really, really helpful. But does your insight get there? Do they use it? Do they find it? Do have do they have access? Are they even aware it exists? That’s a challenge that we see, that many of our customers see, and that’s a challenge where AI really can help and make a difference. But before I talk about AI more, maybe I, first of all, wanna stress the human side of things because we really believe and see that the power of human expertise is irreplaceable, especially here in the context of coral research where the objective is to understand other humans. So you need to connect to them. You need to understand them. You need to find, the subtle differences. And all these things you can only do because you have that shared human experience, because you know what it feels like to be a human, and no AI will ever do that no matter how much you train it. That’s simply impossible. And, also, if you work on the in house side, you know all the context of the organization. There’s all this tested knowledge about what are the hidden assumptions and the hypotheses and what are the strategies. So all of this you can also bring to bear when contextualizing information and making sense of it. Again, a very analog element that is very difficult to translate into technology. So there’s a lot of distinct irreplaceable capabilities and unique ways that that we as the humans contribute here. But still, on the other side, there’s this disconnect we see. Like, the challenge that many of our customers have is while, of course, you have done the research for your specific stakeholder and you work with them closely and inform them and they benefit from it, if it’s not for those specific instances, if that bar is not met to do a distinct specific research in all those day to day moments and decisions, it’s really hard to get access to all those insights for a person on the business side. If you put yourselves into the shoes of them, it’s difficult for them to acquire them, to get them. It takes days or weeks to get responses because, of course, you the insights teams are overloaded and overwhelmed with requests. It’s difficult to get the information in a way that’s meaningful for them and easy to consume and maps to their, problem set. So there’s a lot of hurdles, which de facto means a lot of those insights are not used and the decisions that are made are disconnected from what actually the organization would know somewhere. And here’s some numbers from a survey we did with a partner late last year among, I think, two hundred decision makers and brand and marketing and all the product innovation teams asking how many of the decisions that you do actually use data and insights. And the numbers at this level maybe don’t even look that bad. Two thirds of brand marketing teams, a little more than half of product innovation teams. But if you dive one level deeper, you’ll find that most of the data they will use is rather numbers numbers numbers, transaction data, consumer journey data, etcetera, but not so much of this human element of insight. And that’s a study we’ve done. We’re also very happy to share that if you’re interested, but that really reinforces this point, that there’s a lot of missed opportunity. Imagine all the moments where this human element of insight could help in innovation or brand or marketing or strategy, but where people simply struggle to get to it. And so we have this conflict here. We have all the expertise with the people. We have the knowledge. We have the research, but somehow, for many decisions, we just don’t manage to bring it to the final moment where it can make a difference. And if you think about it as a product in a way, and I think that’s what we need to do, the insights as a product, we have something built, we have something wonderful, and our consumers are actually our internal stakeholders who need that information to make their decisions. If we think about the insights as a product, we might find that maybe we have not enough mental presence with our consumers. They don’t always even think about, oh, we could learn something relevant here from the inside in this moment. Maybe we lack the distribution to get the insights to them when they need them. We also lack maybe the personalization of bringing it in a format and in a context and, illustrating it from an angle, that makes sense to them for what they’re trying to do. So we really think you need we need to think about this whole, endeavor also from a product and customer’s perspective and think about what are the jobs to be done that our internal stakeholders try to accomplish with what we give them. So how can we help them best with the information we have? And that’s exactly where we see AI can help. And, therefore, we can use AI to do a lot of heavy lifting, to alleviate that problem, to be twenty four seven available, to provide access to insights, to provide it in a very low barrier frictionless way. Just go there, interact with the AI, ask you questions, and get the summary of the best that the company knows in terms of all the insights, all the human elements in a way that’s consumable, but also in a way that retains the nuance and retains the link back to all the underlying sources of information and retains the opportunity to dive deeper and then interact with you and immerse themselves. So take away all those routine information requests or even expand on them, make them even more accessible and possible, and at the same time, take away work from the human, element and give ourselves back the time to do the real distinguishing strategic work to do those connections with the consumers to understand what it means to put it into perspective, to interpret it, and then to develop the guidance for the business. So AI here really can be the multiplier for you, the amplifier for the human expertise and knowledge, on the inside side and not what takes away the job. Quite the opposite, actually. So that ultimately is also what we do with our product called DeepSights. Aynsley, briefly introduced it. I would love to show to you how it works. I would love to show to talk to you more about how our customers use it and how they benefit from it. Very happy to do that, off-site, outside the stage here on the exhibition floor. But for now, I want to conclude and just reinforce that, AI, first of all, it’s not a matter it’s not a question of whether we want to embrace AI. AI is here, and today, we use the least AI we will ever use. It will only get more. That’s there. But, also, we saw thirty six percent of respondents said, I expect AI to replace my job. I don’t think that that is a threat here because what we what you try to do is exactly understand the human, and that’s exactly, I would say, the very last domain that the AI will be successful in. But rather, I believe the general mission is to understand how can we use AI to amplify ourselves, how can we use AI to do the tedious things that we don’t necessarily have to do ourselves and free up our time and help us to shine with what really differentiates us as the humans and as what we can do and what we know and to bring that also to our teams. As we heard before, the mission is really to help the teams better understand the user, the consumer, which means, of course, you, we need to have more opportunity to use our expertise to understand, but we also need this multiplication into the organization to bring it to the stakeholders to make it really count where they need it. And that’s where AI can make a big difference. Yeah. And that’s what I wanted to share today. And with that, we wanna conclude and open for any questions you may have. Hello. I’m Royce Yahia from the Allego Group, and we are clients of both, Ipsos and Market Logic. So, I use DeepSights religiously almost every day, and I’ve also trialed Ipsos Facto, and we have our internal, AI, at the LEGO Group as well. So my question is more around the synergies between tools like Ipsos Facto and Insights from Market Logic. Our the DeepSights is great, but it’s not conversational at the moment. Our internal Gen AI is great. It’s conversational, but it’s not pulling data from our DeepSights either. So in terms of your point of view, making DeepSights more conversational or for IpsosFacto, does it pull in data from the DeepSights connection that you have? What are your views in terms of steps forward? Yeah. Great question. Thank you. Maybe if I start, our product at the moment, the way it works, is actually intentionally and deliberately focused on getting a question and producing an answer, and we haven’t endorsed so far this general purpose chatting approach that you would know from Cheggpt because we wanted to make sure we control the experience. We wanted to make sure that people don’t go in and ask silly questions and do silly things and then ultimately go back with silly ideas of what reality would be. So we have very much confined and guided that experience so far, and that has been good. And we had a lot of learnings, but exactly as you say, there’s also limitations with that, and you will see some conversational capabilities coming very soon in the very near future. I’m happy to, talk and maybe show more about that separately. When it comes to integrating with other systems, yeah, we strongly believe the the value of all of this is in bringing together as much information as possible from as many sources as possible. So currently, there is no out of the box integration there, but these are things that we’re looking at always and also depending on what the actual use cases are that customers like you have always are very much committed to making that happen. So if there’s use cases for bringing on more information, be it from partners like Ipsos or be it from other sources, we have a lot of, I would say, willingness and also technical capabilities to make that happen. Yeah. I think that’s a really, really good question actually. And the short answer to it for me is that we’re still figuring that out ourselves. The slightly longer answer is that I think Olaf and his team have done a really excellent job at achieving what they set out to do, which is as he described to actually have a contained, and containable set of, you know, I guess, interface that stops people from doing silly things, and it certainly does that. Ipsos factor is arguably more versatile, but also it brings with it more risks. They’ve, at the moment, we’re using them for two different things as well. So the DeepSights is using it for our internal knowledge management platform, which I run. So that’s got an internal audience. At plus Factor, we we have been trying to market it to clients. So that means there’s different sort of flows of information into it as well. And, yeah, in that in that to your question, we have been curating what goes into DeepSights, which is one of the reasons why, in my personal view, not that of the organization, I think it’s performing better at the moment. And what we’re doing with Ipsos Facto, has you know, we have we’ve had to be a bit more careful about what we put in there because, obviously, our clients are seeing it as well. So that’s been a bit more complicated to figure out. But there is I mean, there’s lots of investment going into that too this year. And, I think it might go quiet for the next few months, but I’m told that in the summertime, again, there’s going to be more releases and it’s going to be remarketed to clients again. So at the moment, I mean, we see out there’s been big news this week as well with, developments in China, and that that’s really disrupted a lot of the financial markets and and lots of big companies that last week seemed completely untouchable are now looking quite vulnerable. So maybe this is the area of the plucky start up we’re about to enter into again. And MarkLogic, I think we’re probably well positioned, I think, in that in that environment. So, yeah. It’s a bit of a weasel answer now, but I think it’s too early to say, actually. Yeah. Still a question on the same table. So I’m Sergio Benavent. I work in Logitech in the consumer insights function. And, what I would like to understand so at Logitech, for example, we are pretty bad at documenting things. So we would do research. We’ll get the decks from suppliers, and we would move on. I think that would be the the situation. And I wanted to understand if, if you would define some I don’t know if there are some basic requisites that you would see that an organization needs to have for being able to embrace the type of solutions that you’re presenting. Yeah. Well, in the times of old, it did require quite a bit of, structural lifting of defining taxonomies, having knowledge management, discipline, etcetera. Now, also, again, AI can do quite a bit of that in the background. So, as long as you have the materials, obviously, in some place, it is very easy to to, load that from wherever it resides on your internal systems and then have AI to roll through that, identify what categories, what topics it is about, and prepare it to be interrogated. So there is very little hurdle to get the basic use case up and running. Yeah. I mean, I I I’m a knowledge manager by trade, or at least have been for the last fifteen years, and I had it beaten into me at knowledge management school. Metadata, metadata, metadata. That was literally, you know, the the it was the alpha and omega of the of the discipline. And when AI started to emerge, I think some people, not old enough, not the people who actually understand the technology, but maybe some of the business users of it were delighted because I thought, oh, we don’t need to get get rid of all that tedious knowledge management stuff. We can just here’s a silver bullet. We can just throw our stuff into it and it will magically produce, the answers that we want. And I think, you know, we have some really good data scientists working at Ipsos. And, I am reassured by what they’re telling me and by my own experience of working with MachLogic and Insights that some of these old things that the old disciplines and practices remain just as important as ever. I mean, it’s not the magic you know, it won’t magically solve all of your problems. You still need to do some of the, I think, the fundamentals around information governance and tagging and and and that kind of thing. It will it will improve the experience that, that AI will give you. At least that’s been my experience. I don’t know. I don’t know if you think I’m being a bit of an old footy duddy or not. Oh, you you have to say that. Thank you. No. For sure. There are elements there. Perfect. Thank you, Olaf and Ensley, for that great session.