the risk you’ve assigned to that particular player. So how are you going to do all that if you don’t have computers and AI to assist you? So I believe that now we’re no longer talking about whether or when AI is coming: It’s here and you can’t work without it. IH: Earl, you had a comment EH: I don’t like to talk about what I do in my day job just out of respect, but I do have a small story. 15 years ago, I was asked by an investment group to look at the future and we didn’t look at this industry, we looked at the general tendencies. Because Facebook was a couple of years old, people were starting to buy a lot online with their credit cards. And we were starting to create avatars at that time, this is around 2011, we were starting to create avatars of human behavior, which is my real passion in life. So there were three easy predictions you could make about 15 years ago: one, the world was going to cloud computing; two, the world was going cashless, and three, the world was going to be completely AI driven. This is 15 years ago, just log on to Gartner Group and go back and have a look. So how will this work in the land-based gaming industry where most of the data is sitting under tables on desktops with dust in them. They’re in proprietary languages, and they’re impossible to extract and consume. So can you imagine the technology debt that the land-based gaming industry is accumulating at an increasing rate right now? I just don’t understand why our industry is not 100% wall-to-wall KYC. If you have nothing to hide, you should be able to play. Why? Because once you’re mining the data, money laundering, and addiction disappear. Our cloud database has 57 billion lines of data from 12 years of gathering it from 45 countries. We’ve whittled it down to about 11 triggers that are the genesis of addiction. Here’s the cool thing. They’re also the same triggers for money laundering and they’re the same triggers for marketing. Once we arrived at that realization we gave all of our data to UNLV and their AI department is right now eating it up and doing whatever they want to do with it. But the point here is there is no difference between player protection and player marketing. It’s the same fundamental data that we’ve got to consume if we’re going to increase our revenue and fight off the other entertainment segments and protect the player. RK: I absolutely agree with you, Joseph, that the idea that a human being could digest and analyze in a meaningful way the amount of data we see with gambling operators is kind of a bit outdated. But still, we see a lot of that happening when it comes to the RG data analysis and detection. And I also agree with you Earl that that if we could have the data from the land-based side, which has up until recently been a challenge, but actually we are moving into pilots where we are able to get the data, the challenge. Corrupted, potentially, yes. But, we work with the operator to fix that because they also want to be able to do detection of the gambling behavior for player protection purposes, from the land-based side as well, just as they already do online. And thanks to ChatGPT, it has become much easier to check the conversation with operators. Because there is a bigger understanding that AI is something like the internet: It has come and it will stay for the foreseeable future. In fact, ChatGPT is probably the lowest level of AI. IH: Let’s jump into the AI algorithms. Outcomes are only as good as the data and the inputs. You know the saying: garbage in garbage out. I’ve been working with data and systems for close to 30 years. It’s gotten better but collecting data is still problematic, because there’s a lot of different devices out there, they communicate in different ways. So if we look at use cases, we talked about Responsible Gaming, one of the areas that I’m working a lot on at the moment is AML, a big issue in Australia so we’re asking what extra information needs to be generated for the land-based side to be as effective at AML and RG issues. Some of those have been identified. The New South Wales Crime Commission published its report, the Islington report, and I worked with them a little bit on the data sets. What we found is one of the problems was that the data was only collected every 15 minutes. It was not real time. It was anonymous, it wasn’t associated with player information. What are some of the additional data sets and quality of the data that is required into that infrastructure for AI to be truly effective, both in the RG and AML spaces. EH: So now you’re in my direct wheelhouse. This is going to shock you. I don’t want to insult any manufacturers in the room. But once again, we connect to 3000 different machine types in 45 countries. There is not one machine that we have met, that when you plug in the SAS cable that the machine streams all of the SAS channels, not one in
59
Made with FlippingBook flipbook maker