Hi, and welcome to the Three Questions podcast with me, Lianne Johnston, the founder of Affiliate Insider. And today I'm super excited to join this podcast with Kim Moritz from Mindway AI. Hi Kim, how are you?
SPEAKER_00Hey, Leanne, good to meet you. I'm fine.
SPEAKER_01Thanks very much for joining us today. You've got a really interesting topic to talk about. So I'd like to kick off with just asking you to explain a little bit about Mindway AI and your background and how you came to actually develop this product.
SPEAKER_00Thank you. Yes, very happy to do this and uh thank you for inviting me. Um, so uh if I start perhaps with my uh my own background, um I'm a uh full professor in uh neuroinformatics here in uh in Aarhus, and um I have spent a lot of time uh on sort of the uh the merging between statistics, artificial intelligence, uh, and then neuroimiting and neuroscience. And what I find you know really interesting is to consider how we can take many of the research results, uh the research knowledge and know-how that we uh gain over time, and then get that you know out to society building tools that you know in this case uh at the end of the day, uh, you know, we hope will benefit the customers in in the gambling industry. And um, so in particular, uh what I've been interested in is to bring results from from neuroscience and neuroimaging uh into this field. We um started a long time ago uh actually with uh with Danske Spil here in Denmark to consider uh whether we could use in that sense artificial intelligence combined with uh neuroscience to identify uh customers on a trajectory to becoming uh you know uh at-risk uh gamblers or problem gamblers. And um then that has evolved, and and I hope to get to talk a little bit about that because now we're at a point where we are not only uh detecting uh problem gamblers uh with you know with online operators and but also with uh brick and mortar casinos uh because we have developed uh certain tests that can be you know seen as complements to the the PDSI, the problem gambling severity index. And we are also now uh sort of uh the state of the art that we are doing right now, we are considering whether you can also offer online sort of um you know training tools uh not to you know limit people in in their access to um to uh to gambling resources. Um we're not talking about uh you know exclusion methods, but talking about you know how can we essentially uh help people become perhaps less impulsive if if that has become an issue and people have a tendency to not be able to stop gambling when they actually want to.
SPEAKER_01So it's actually really nice because it's a pre-detection um mechanism that people can use to sort of spot certain behaviors, but then it also is a detection method as well, an early detection method, which operators can then use to um, you know, uh do player profiling to prevent anything further from happening. So it would you say it's more of a preventative tool than an engagement tool that's prior to the problem occurring or becoming a problem?
SPEAKER_00Um we we believe a lot in in uh prevention and if not prevention, then at least early detection. And um, you know, we have um from from uh machine learning, from uh deep learning and so on, we have uh very strong technologies that uh we have seen a lot of success with, uh, for instance in in the clinical field. So essentially trying to automate uh the processes that let's say a radiologist need to go through in in reading complex uh medical data. Yeah, so there's a lot of strength there, and and what we sort of asked ourselves is whether we can sort of bring that into uh to gambling, because um that's the sort of the same uh issue that we have very detailed, uh almost spin-level um data available. Uh and yet, you know, it's been a problem also for many operators to sort of really um exploit that technology. And I think one of the problems is that um you know we haven't had a good opportunity to have a solid target. So you could say that, you know, in sort of the medical field where I also uh come from, you know, if you read a radiological uh you know MR image, you you decide whether you see pathology or not. Uh you know, if you're Google and and you you want to do image classification, then you take relatively complex data such as uh such as a high resolution image and you can put that through a neural network. And then because the target is so clear so everybody can see you know whether you know these this image is a cabin or a dog, we can build very strong technologies to essentially mimic uh what you know the human or or for the radiologist or the expert can do. But the question then is in you know in in gambling, what is it we are predicting? And I have seen you know a lot of efforts where what people really want to try to to uh obtain is to use artificial intelligence to predict um you know people at risk of self-excluding.
SPEAKER_01Yes.
SPEAKER_00And you know, we we see that as sort of a uh even though it seems you know a clear target, one of the problems is that we we know from from earlier studies that it's only around 20% of uh people who admit to have a problem that actually or who actually use self-exclusion. And also the other way around, uh we also know that it's it's roughly 30% of people who use self-exclusion actually at the end of the day perceive themselves as problem gamblers. And that means that you could say sort of the sensitivity and specificity of self-exclusion as a method, um, you know, as a proxy marker of problem gambling is not uh so well suited. And that means that you know we could have essentially the best AI uh algorithms in in the world, but if what we're you know shooting for is not really discovering you know at-risk customers, then you know we're not really utilizing this technology well.
SPEAKER_02Okay, that's right.
SPEAKER_00And that's kind of where we uh you know said from the beginning, you know, let's see if we can learn something from you know from the research, from the clinics and so on. So what we're essentially doing is that we are asking uh you know a panel of experts that that we are working with, and these people are uh you know psychologists, they are researchers, um, uh we even have uh medical professionals. And these you know uh highly regarded experts are used to working with uh you know problem gamblers and and and customers who end up seeking help uh for any uh problem they they might have. And therefore, we feel that you know if we um actually show uh a bit of data um for particular customers to these experts, then they are kind of the best um you know uh post in the world to make an assessment whether they see uh a problem emerging or already manifest for a particular customer. So it's much like you know when you go to your general um physician and you could have a complaint about a stomachache. Um, you know, it's it's not you know as a necessarily a simple test, but you know, the physician would have sort of a gut feeling and a holistic view on on whether this is something serious or not.
SPEAKER_01Yeah, there's like a level of diagnostics that happens before you come to the root result. But the thing is is that you need, and I guess what this tool does is it tells you you've got a stomachache, you need to go see someone. So it's the prevent it's the prevention, and then obviously you you get the support that you need, or um, you know, and then and then go for the counselling and and all the rest of it. Now, the thing that's interesting to me is that data is very regularly available in in the eye gaming industry, and obviously operators are much more concerned about um, you know, safer uh marketing practices and making sure that they're marketing responsibly, and so are affiliates. Um, so you know, how how do you actually want operators to be working with you? I mean, is there is it like a plug-in that they would stick into their back-end systems that then fires off alerts, which could then trigger a email communication to you know the players as they segment their databases? I mean, how's what's the practical implementation of of the product and how are you hoping to work with uh you know marketers and games um managers or casino managers or sports book operators that deal with customers end-to-end? I mean, I noticed on your website you've got this really great white paper, so I just want to call out and say anybody that's interested in in reading more about the um research research that you've done, if you visit mindware-ai.com, you can request uh the white paper that that's been developed just by completing the simple form here. Um but moving on from that, you know, how do you envisage the industry will actually you know accept and use this technology when it's when it's ready to go live?
SPEAKER_00Yeah, that's an that's an excellent question. So what it is at at the end of the day, it's a software, and you can imagine that uh you you can think of this software as uh basically a virtual expert or a virtual board of experts. So essentially what it does is that it sits um with the operator. Importantly, you know, it's it's not a cloud solution, so we're not sending uh or transmitting data uh solution that sits with the uh within the uh operator uh firewall. And it takes uh on a routine uh basis a sample of data uh acquired uh over the past period of time, and then it makes an assessment, and that assessment we know and we have shown to to a great accuracy, then mimics that out of the LXPs. And then the um as sort of an additional uh you know advantage, uh we do not only say that you know these customers might have a problem, of course. By far most customers come out you know with with a perfectly harmless behavior. Yeah, but for the things where you know we do see uh developing signs of uh problems, then we can use sort of the uh intersection of the AI and then the psychological expertise to actually devise um conversation guides, if you will, um for the operator to use in their dialogue with the customer. And that's something in in our experience that uh you know ends up resonating well uh with customers because it means that they uh perhaps uh feel that they are met uh in a more uh you know on a on a more informed basis, they might better recognize you know developing problems, and that that's really the basis for um for intervention. And I think you know, one one of the places that you can then go from there, um, you know, talking also about you know how can uh affiliate and operators uh use this, uh I mean, one thing is to detect that somebody might have a problem, uh, another thing is to actually reach out uh and have a direct dialogue, but but it might not even come to that. And and I think you know, many uh operators and and customers for for really good reasons would prefer that we would catch these problems much uh earlier before we can talk about a problem. Yeah, and one thing we are doing there is that um you know, if an operator would see that here might be a sign of of uh problematic behavior, they could, for instance, point them to a kind of a self-test that we have created. Um, it's um a rather different uh self-test than what's on the market today, which are primarily based on you know a verbal uh you know uh questionnaire. Yeah, so we'd be asked in the PDSI or the game test, you know, if you feel in control of gambling, if you lost you know a lot of money, you know, if you have borrowed money to gamble for and so on. And and those are really you know uncomfortable and and rather you know private questions. Yes. It's like asking somebody uh how much they they drink. You know, we all know what the what the best answer is, and perhaps that's that's what we're getting. Uh, what we're suggesting instead is uh if we could actually engage customers in playing a game uh which is well understood by us and by the research community, then instead of asking people how they do perform and how they do make decisions uh when they are gambling, then we can actually observe it. And what that means is that uh we can also provide feedback to the customer, which is uh you know based again on psychology and artificial intelligence, where we can say, you know, what what you have, you know, the the way that you have been um you know behaving on this particular and it it's it's a rather short game, but uh you know what you're doing right here and what we believe could be a force of concern is something we can then tell pretty directly uh to the customer. And in that sense, you know, we uh we have something here that is not you know one-dimensional in in the sense that the PDSI, for instance, would say, you know, uh you have a score of eight uh on this uh on this scale, which doesn't really help the customer, but we can say, you know, in your behavior when you're making decisions on on bidding, then you know, we feel that that this is uh an advantage, and we feel that uh some other things that are going on in your gambling could be a source of problems later on.
SPEAKER_01Great, thanks very much for sharing that with us. So the last question in our three question podcast is um I wanted to just bring up the fact that in recent news we saw that uh quite a large affiliate company has actually invested in in helping the business to grow. Um, Better Collective is one of the leading sports betting affiliates in the iGaming industry. Um, how has that partnership come about and what's in store for the two of you to be working more collaboratively together in the future as this product grows?
SPEAKER_00Yes, let me first say that this is a partnership that obviously we are tremendously uh happy about and and that uh you know is already really uh fruitful. Um we we believe that you know, with uh the partner we have here, you know, we we are really sharing uh values and and we are sharing sort of a a vision for where we want to go uh with prevention and early detection of ending. It it was a dialogue that was started uh about one to one and a half a year ago, and and we have had uh very close dialogues uh ever since and and actually had a uh a very good process moving to um to where we are today. And you know, for us is in mind we are as a you know a basically a startup company, uh we uh need uh you know to be uh exposed to the market.
SPEAKER_02Yes.
SPEAKER_00And and we have certainly you know uh done that on on our own, but of course what we see with uh better collective is that we can uh significantly uh increase uh that that bandwidth.
SPEAKER_02Yeah.
SPEAKER_00At the same time, you know, we we are hoping that uh better collective in in us see a uh a partner you know uh who can strengthen uh also their ambitions in uh in responsible gaming. And uh what we're seeing is that that this is being a very successful uh collaboration.
SPEAKER_01And also, I mean, often affiliates are kind of left out of the equation when we talk about you know um you know awareness about problem gambling, it's always directed towards operators and the play. You know, the message is always directed there. So it's quite interesting to see that an affiliate group is being progressive and actually wanting to get involved in um supporting you know gambling addiction and and making sure that you know best practices are put in place by using technologies such as yourselves. So just to close off our podcast, thank you very much for sharing all of this information and detailing exactly how the product works. If there are any operators that are listening to this podcast and would like to get in touch um with you, how would they get go about doing that to actually find out more about the service?
SPEAKER_00Oh, so we we'd be happy uh if people contact us uh directly. Uh so we have a uh webpage, uh mindway uh.com, where you can certainly find much more information. But of course, we'd also be very happy to get directly in contact, and you are very welcome to contact me uh on my email address, which you can also find in there. But it's Kim Kim at mindway.com, and we'd be very happy to to engage in dialogues.
SPEAKER_01Brilliant. Thank you so much for joining us on our podcast today, and we look forward to seeing how Mindway expands in the future.
SPEAKER_00Thank you, Ian. It's been a pleasure. Thank you.