As artificial intelligence becomes an increasingly essential tool for gaming operators, there are more discussions about how best to use the technology. What’s important, according to OPTX Vice President of Data Science Steve Bright, is to differentiate how AI is used in the industry versus general uses of ChatGPT and other artificial intelligence tools.
“We’re really focusing on leveraging machine learning and AI to solve some domain-specific problems,” Bright says.
The Las Vegas-based company was incorporating AI in its products before the recent surge in widespread usage. OPTX uses artificial intelligence to help clients maximize slots, marketing and player development performance.
Steve Bright
But while the use of the technology continues to increase, Bright emphasizes that the company uses AI as a “co-pilot, not an autopilot.”
“Everything we do with OPTX AI features is intended to supplement decision making, not replace it,” he says. “What we’re looking to do is offload some of the, not necessarily tedious, but some of the repetitive point-and-click, cut-and-paste, write a custom spreadsheet things. … We’re trying to offload that to an algorithm. We want to free up the people at the casino, the people at these properties, for actual human interaction. Because this is ultimately about the managing and cultivating and curating of fun player experiences.”
The idea is that if managers are able spend less time poring over spreadsheets and data, they can spend more time on casino floors meeting and greeting customers, providing a warmer, more personable experience.
But there is a drawback to relying too much on artificial intelligence. Bright says data scientists who have worked with AI will admit that the technology’s range of accuracy is between 70-90 percent. Which means that AI can make errors up to 30 percent of the time it’s used, though this depends greatly on the nature of the problem under consideration – and where we have had data for comparison, we have found that the AI is more accurate than the human.”
“We want to measure everything,” Bright says. “I’m really trying to bring a scientific approach to problem solving. To my mind, that’s what data science is about. We’re all about applying the scientific method to business problems that are using data. We always want to know the impact of every AI recommendation. We want to know what happened afterward.
“Ultimately, we’re making a prediction. If you change this game on this machine, this is what we think will happen. If you increase this player’s offer, this is what we think will happen. And whenever you make a prediction like that, as a scientist, you probably want to ask the follow-up question: It’s now 20 days after I made that prediction, what actually happened here? Was my prediction right or wrong, or more precisely, how wrong was my prediction?”
OPTX has three products that can provide artificial intelligence-based solutions: Marketing AI, Player Development AI, and Slots AI. The products can be deployed to fit the needs of individual operators to target specific groups.
For instance, Bright notes that about half of the revenue at a property comes from 1-2 percent of its players’ base. For those players, AI determines patterns and preferences.
“It’s giving the human host some insights that they can refer to when they are having conversations with those individuals, making the personal touch more personal,” Bright says.
For players not quite on that level, but who still come to a property frequently, it’s possible to develop curated experiences by determining optimum offer amounts and gaming preferences.
“Those are things where you don’t necessarily need a human to directly manage the one-on-one conversations,” Bright says. “You might have a human doing a final check of a campaign creation, or a final check of an offer. Basically, AI can automate some of the communications and fine tune the offer and really understand player preference to effectively act as a casino host in an automated way.”
Bright adds that OPTX is exploring the incorporation of more general AI tools, such as ChatGPT, into its development process and product features. But the current emphasis is on production features based on domain specific custom AI models.
