Conference proceedings, Zurich 2023

and say was the inflection point. So artificial intelligence is not something that is coming: it’s something that’s being adopted at a speed no other technology has been able to because it took 25 years to go from mainframe to personal computers to cloud because the food that AI eats, drinks and breed’s is data. So we’re in an inflection point right now that is not going to only affect players, it’s going to affect the human condition from every single point of view. IH: Earl, you touched on ChatGPT, which most people are aware is the generative form of AI, it’s one of the subsets of AI. So let’s start talking about AI. AI has a fairly broad definition. Earl touched on a specific use case but one of the definitions I like to use is the ability of machines and systems to acquire and apply knowledge and carry out intelligent behavior. That’s one of the definitions, I think that’s a UN definition or an OECD definition, fairly broad. When we talk about AI, it’s not something new: AI has been in place since 1966 or 1967. Some of the first large language models emerge around then, but it’s only because of modern computing and the accessibility of the powerful computing devices that we all carry with us that it’s really exploded into that point of inflection. So Rasmus, turning to you, what are some of the things that you’re seeing? How is AI going to impact or change the way in which players and operators and manufacturers work within the gaming industry? RK: Thank you very much. We at Mindway AI have invented the opportunity to utilize AI to a far more individualized, earlier, more granular detection of at risk and problem gambling. From our scientific background, coming from our university in Denmark, where we combine machine learning algorithms, which is another discipline different from language models like ChatGPT. But what we have found in the research behind the innovation we have made, was that AI is really capable to find patterns and conduct big data analysis. The challenge when it comes to at-risk and problem gambling behavior, or analyzing really any kind of human behavior, is that we are talking about individual people here. So our human behavior is unique, and so would our gambling behavior be. What we have invented is the ability to use human assessment from psychologists, and recently also lived experience, and built that into the training of the algorithm. So we replicate psychologists assessment when the algorithm is run. We take the best from the human world combining it into the AI world and use the human assessment to achieve quite a high accuracy with the algorithm. We have taken the initiative to get our software tested and validated by GLI, and they found that in at least 87% of all cases, our algorithm detects the same as a human psychologist when it comes to the gambling behavior. Bearing in mind that no human psychologist is 100% accurate at detecting either. We don’t do anything else except problem gambling solutions so we strive improve every day. And so far, our algorithm currently detects roughly seven million active players a month in 61 jurisdictions. IH: So one of the use cases that you’re saying, Rasmus, is that AI has a role to play in identification of patterns of behavior which might indicate that a player is at risk. RK: Absolutely, and even detecting behavior before it is considered problematic because we have risk factors, indicators that could indicate a customer is moving into behavior that could be problematic later on. So then, of course, it’s a matter of how early the operator wants to flag this to their customer. Do they want to wait until it’s something that is to be handled as an intervention? Or do they want to do it earlier, even just as a kind of caring response from a commercial company to their customers? If they do the latter, then there is an opportunity that the customer will stay on a more sustainable level, which actually, in the long run is potentially a better business case for the operator. IH: Great. And, Joseph, would you like to add anything to that? JB: Well, I’d like to support Earl’s point that if you’re not using ChatGPT today, it’s like you’re using a fax machine rather than email. I subscribe to that and I’ll add that actually the use of AI today is no longer a nice to have, it’s a necessity because of the amount, the magnitude of data that has to be processed. So if you look at online gambling and the obligations they have in terms of AML, in terms of Responsible Gaming, in most jurisdictions, how can humans process all the data? The only way to process that data is through AI. And this will be much faster, much cheaper. But it’s not just about saving money. In Europe, the moment that a player hits €2000 in transactions and deposits, that person has to be KYCd and a risk factor applied, and you have to actually start monitoring the transactions according to

58

Made with FlippingBook flipbook maker