Can Artificial Intelligence (A.I.) Assist Casinos in Reducing Problem Gambling?
This article is the first in a new series on the potential for artificial intelligence to tackle common issues.
Alan Feldman strolled into the exhibition hall of ICE London, a big gaming industry event of Bahis Siteleri, a few years ago.
Mr. Feldman spent the previous three decades as an executive with MGM Resorts International, where he focused on compulsive gambling and its financial, psychological, and professional consequences. Prior to his departure from the firm, he assisted in the development of a national responsible-gambling program aimed at assisting players in changing their behaviour in order to lower their risk of becoming problem gamblers.
While on the ICE floor, he spotted many businesses pitching innovative solutions that would utilize artificial intelligence to not only detect but also forecast problem gambling. Mr. Feldman expressed quick skepticism. He considered how A.I. might be used in a variety of ways, but he had never heard of one that anticipated a state of mind.
Artificial intelligence as a panacea for compulsive gambling "raised many more concerns than it answered," according to Mr. Feldman, who is now a distinguished fellow in responsible gaming at the University of Nevada, Las Vegas' International Gaming Institute. "It was polished, intriguing, and intellectually fascinating. However, whether it would really do anything, I believed, remained very much in doubt."
Another apparent question: Isn't a problem gambler precisely what a casino wants financially? Mr. Feldman's response was succinct: no. Even if regulatory concerns are disregarded – gaming operators face fines or license revocation if they fail to monitor problem gambling and intervene when required - it is, counterintuitively, not in their financial best interests.
"Casinos need clients in order to survive," Mr. Feldman said. "And the only way to have customers is to have healthy, flourishing consumers who are able to pay their bills and return the following time." Gamblers with a problem "usually end up the same way," he noted. "The final result is the same for all of them: they lack money."
Not everyone agreed. According to academics who research responsible gambling, such initiatives are tasked with assisting individuals without impairing general operations.
"Some operators have more comprehensive responsible gambling programs than others," according to Lia Nower, director of Rutgers University's Center for Gambling Studies. "However, there is a profit motive at work, and I have yet to see an operator in the United States invest the same amount of money and effort in developing a system for identifying and assisting at-risk players as they do in developing artificial intelligence technologies for marketing or extending credit to entice players to return."
In any case, the combination of artificial intelligence and gambling makes perfect sense: unending and consistent data; decision-making; automated systems. With the rise of internet gaming, the possibilities for harnessing this combination for the greater good seem limitless. The reality is significantly more difficult – analyzing human behavior, managing privacy regulations, and resolving regulatory obstacles.
Simultaneously with Mr. Feldman's reservations about those prospective remedies, Danish academics were attempting to resolve the same issues. Mindway AI, a spinoff from Aarhus University, performs just what Mr. Feldman questioned: it forecasts potential problem gambling. Founded on research conducted at Aarhus University by its creator, Kim Mouridsen, the firm trains A.I. algorithms to recognize behaviors linked with problem gambling via the use of psychologists.
One important obstacle is that there is no one signal of whether someone is a problem gambler, according to Mindway CEO Rasmus Kjrgaard. And in the majority of casinos, human identification of problem gambling is limited to a few parameters — mostly the amount of money spent and the amount of time spent. Mindway's approach considers 14 distinct threats. These include not just money and time, but also canceled bank withdrawals, changes in the player's playing time of day, and irregular wager fluctuations. Each component is assigned a value between 1 and 100, and the A.I. then develops a risk estimate for each player, increasing with each hand of poker or roulette wheel spin. Players are graded on a scale of green to crimson red (immediately step away from the game).
To customize the algorithm for a new casino or internet operator, Mindway provides its data to a team of specialists and psychologists specialized in spotting such behavior. (The corporation said that they were compensated independent consultants.) They conduct customer assessments for each client and utilize that model as a type of baseline. After that, the algorithm repeats its diagnostic throughout the whole client database.
"As soon as a player profile or player conduct changes from green to yellow and then to the other phases, we may intervene," Mr. Kjrgaard said. The program's significance extends beyond detecting blood-red problem gamblers; by tracking changes in Mindway's color spectrum, it forecasts and catches players as their play devolves. At the moment, he said, casinos and internet operators concentrate their efforts on blood-red gamblers; with Mindway, they can detect gamers before they reach that stage.
According to Brett Abarbanel, head of research at U.N.L.V.'s International Gaming Institute, the most difficult step is taking that data and conveying it to a player.
"If my system identifies someone as a potential problem gambler, I'm not going to send them an email saying, 'Hey, fantastic news! My algorithm has identified you as a possible problem gambler.' You should immediately cease gaming!'" Dr. Abarbanel's statement would be self-evident: "That is what will happen."
How to transmit such information effectively — and what to tell the gambler — is a point of contention. Some online gaming firms communicate through pop-up windows; others communicate via SMS or emails. Mr. Kjrgaard expects that customers would take his data and, depending on the amount of risk, contact the player directly through phone; the uniqueness of the data, he said, enables such calls to be more personalized.
Since its inception in 2018, Mindway has contracted with seven Danish operators, two in Germany and two in the Netherlands, one worldwide operator, and a sports betting operator in the United States, Mr. Kjrgaard said. According to the firms' annual reports, Flutter Entertainment and Entain have also worked with Mindway.
Because this technology is so new and there is no regulating organization establishing standards, Mindway and related firms are largely operating on their own at the moment. "We wanted to be able to communicate to you and anybody else — operators, certainly — that we not only supply scientifically sound software, but also want a third party to validate our work," Mr. Kjrgaard said. "However, it is paradoxical that there are no particular standards that I can direct my staff to meet."
At the moment, Mindway's technology is mostly used in online gaming. Operators integrate Mindway's GameScanner technology into their portal, which assesses not just individual threats but also the system's overall risk. Applying that degree of supervision to in-person gambling is far more challenging.
Macau is one example of a measure of success. Casino operators there monitor players' betting activity using concealed cameras and face recognition technology, as well as poker chips equipped with radio-frequency identification technology and sensors on baccarat tables. This data is then sent to a central database, where the performance of each player is recorded and interplayer collusion is monitored.
This is the future, Mr. Kjrgaard stated: financial incentives will drive achievement. "Smart tables" and attempts to combat money laundering and comply with financial rules may someday provide the data necessary to accelerate the use of artificial intelligence to in-person gaming.
(It also indicates another barrier to using artificial intelligence to gambling: cultural differences. Players at Chinese casinos, Dr. Abarbanel said, are used to this amount of surveillance; this is not the case in the United States.)
Mr. Feldman noted that although artificial intelligence will undoubtedly benefit casinos in terms of marketing, promotions, and game selection, he remained wary about its application to assist problem gamblers, despite recent advancements. He feels that such a tool would be more effective if utilized individually rather than widely, similar to the "Your spending is 25% greater than last month" warnings that crop up in online banking accounts.
"It's similar to drinking. Is there anybody you know who has never been drunk? That does not always indicate they are an alcoholic," he said. "However, that one drink every night that has morphed into one and a half, maybe two, occasionally three — perhaps you want to rein it in a little bit. However, you do not want the bar monitoring every record here, correct?"