G2E: Panelists weigh benefits, risks of AI

Wondr Nation Vice President of Marketing, Strategy & Operations Dyani Marvel promised Global Gaming Expo attendees “a very exciting discussion” of the intersection between artificial intelligence and responsible gambling.

Indeed, the panel session “Harnessing AI for Enhanced Customer Relationships and Responsible Gaming” was frequently punctuated with dramatic thunderclaps from the skies above the Venetian Expo Center.

Shelley White, CEO of Canada’s Responsible Gambling Council, boiled the topic down to, “It’s really about creating safer environments for the players.” She was joined in the discussion by Kasra Ghaharian, director of research for the International Gaming Institute at the University of Nevada Las Vegas. Rounding out the panel was Michael Auer, a psychologist and statistician who also works for Open Bet. Auer said he’s been researching player data and responsible-gaming tools since 2008 and has subsequently published on the topic. When asked where attendees could learn more about AI, he jokingly replied, “Read my papers.”

Auer explained that AI is typically known as “advanced algorithms,” the term he prefers, as definitions of AI vary. Advanced algorithms are used to identify players who show patterns of disordered behavior.

Ghaharian cautioned that there were still risks inherent in AI, particularly in the form of biased data that could lead to gender or racial discrimination. But there are risks in not proactively deploying AI, particularly with regard to fostering AI literacy. “If stakeholders don’t know what they’re talking about, operators aren’t going to make good decisions. There’s a risk if we start to develop these things behind closed doors. How good does this thing actually work? Because there’s not a lot of information.”

Ghaharian contrasted the closed-loop nature of Open AI with the more transparent nature of Meta, which he prefers.

“These are topics very much in discussion these days,” interjected White. AI, she said, “has the potential to be used very effectively in terms of early detection and positive customer interactions.” By the same token, unlicensed operators could employ AI to wreak harm.

“We’re just scratching the surface,” Ghaharian resumed. “I don’t think we know what we’re talking about,” particularly with how generative AI can be used to implement responsible-gambling tools. Formulating complex methods, he said, takes time.

Marvel asked, “Can AI personalize player responses?”

Personalized interaction works very well, responded Auer, so why not in responsible gaming? “It’s done very very rarely. It’s very hard to actually do it.”

However, Auer said that studies have shown players rarely withdraw their winnings, but usually re-wager them. “If you nudge players [with AI] at the right time, it works very well.” As an example, he suggested messaging players when their winnings are bumped upward. AI, he said, could handle a huge volume of such interactions.

White cited Italy, Ontario, New Jersey, and Colorado as jurisdictions that are ahead of the curve in requiring use of advanced technology for responsible-gambling efforts. She said she was looking at how AI can be used in marketing to tailor sensitive messaging for players.

She also raised the question of what kind of language the sites are using to talk to players, based on previous play. She also stressed examining the intersection of human and AI interventions.

Marvel raised the matter of the SAFE Bet Act, recently introduced on Capitol Hill by Sen. Richard Blumenthal and Rep. Paul Tonko. She explained that AI is addressed in the bill, including prohibiting its use to observe players’ habits, interactions with players, and tracking microbetting.

“I don’t see a world where that’s going to go through,” said Ghaharian, dismissively. “I worry about the vagueness of ‘AI’ there. There’s this kind of murkiness around AI. Just an outright ban on the commercial use of AI, I don’t think that’s going to fly. It’s just a bit of hype now.”

White segued into an observation about insufficient dialogue between operators and governments of the sort that would lead to common standards and practices. Yet, “this is something that’s being talked about globally.”

In Europe, Auer noted, several jurisdictions require companies to monitor player behavior for indications of disordered gambling. A lack of transparency, he warned, hinders intelligent implementation. “The biggest issue is a trust issue,” he observed.

Some regulators, he related, audit very specifically, in terms such as, “Why did you allow that person to gamble for three hours?” “AI doesn’t work like that,” he noted. “We’re at an evolutionary stage and it’s absolutely imperative that we assign appropriate resources to ensure that there’s ongoing research to measure the efficacy.”

Ghaharian cautioned, “It’s going to be very interesting in five years when we have artificial intelligence that’s as intelligent as humans. I don’t think it’s science fiction at this point.”

Bringing it back to gambling, Marvel observed, “When we have problem players, that doesn’t help them and it doesn’t help us.”

This led Auer to point to a disconnect between AI and human intelligence. “AI is based on logic and humans aren’t not logical,” he quipped, likening AI to the characters of Spock and Data on “Star Trek.”

Asked for a closing sentiment, White supplied one, “It’s imperative that we optimize the potential of AI for good.”