by Mateusz Mazur

“There is a lot of similarity between investing and being a professional poker player”

In another exclusive interview, we have the privilege of speaking with Mayank Kejriwal, a Research Team Leader and Research Assistant Professor at the University of Southern California. He also recently took on the role of moderator at the SBC Summit where delved into the potential impacts of ChatGPT and generative AI technologies in various sectors, including the gambling industry.

Can you tell us more about your recent involvement as a moderator at the SBC summit regarding ChatGPT and generative AI’s potential impact on the gambling industry?

I was invited to moderate a 45-minute panel on the influence and impact of generative AI technologies like ChatGPT on SEO and web search, with specific focus on the gambling industry. We had three panelists with a diverse set of experiences and viewpoints.

We addressed questions ranging from (potential) automation of SEO professionals, to whether ChatGPT will be a net benefit for users. We closed the panel with statements on where the technology is likely headed in the next few years, and resources that audiences should follow to learn more about these technologies as they evolve. Moderating the panel was a rewarding experience, and I believe the audience benefited from our discussion as well. We had a great turnout, overall.

In what ways do you believe ChatGPT and generative AI could influence SEO or web search within the gambling industry, based on your panel discussion?

Generative AI is going to change the way that SEO and web search will be done, but it will not necessarily ‘automate away’ the human element. ChatGPT democratizes technology in a way that is unprecedented. Now, anyone can ask ChatGPT to generate HTML snippets, or to point them to SEO resources.

Therefore, the real barrier is not technological knowhow but creativity and content. SEO professionals and marketers will have to wear the hats of content creation and generation in a way to make their websites and digital offerings stand out. I believe this is a net benefit for the industry, and my hope is that it will elevate the quality and findability of content on the Web.

Could you share some insights into how academic work in fields like game theory and decision-making can contribute to the online gambling industry?

John Von Neumann was famously inspired by poker when he devised what we today refer to as game theory. Poker puts game theory in practice: how can we best make decisions under conditions of uncertainty and hidden information? Online gamblers are constantly trying to get an edge, including by doing second-order thinking (does my opponent know that I know he knows?). Game theory helps to operationalize that in a mathematical way and can be a powerful tool for gamers.

The science of decision making has also become very important to games like poker in recent years. We now know that humans are fallible and biased, especially when they are stressed at the poker table. People overweight recent events too heavily, and sunk cost fallacy is still difficult for many to overcome. By having a better understanding of these biases, online gamblers can better control their impulse and make optimal decisions even under stress. It is perhaps not surprising that some former poker players have gone on to write books on decision making and strategy. Therefore, for online gamblers and poker players looking for an edge, they should look into this work!

What about the potential use of advanced AI in training poker players and building AI for gambling. How feasible is this, and what advantages do you see in applying AI in such contexts?

Since the rise of cheap computational power, professional poker players today rely heavily on simulation software and advanced modeling. A key element that has been missing is the modeling of an opponent’s ‘psychology’. Simulators do allow one to capture a player’s risk preferences through variables like opening card ranges, betting frequency, and so on, but the actual ‘thought process’ or even personality of a player, which is essential to winning games against real human beings, has not been captured properly in software.

Advanced AI that is coming out today may allow poker players to fill this gap and lead to even more competition in the professional circuit. It may even be possible for near-future generative AI to automatically train AI bots that simulate a given poker player, just by watching YouTube videos. Today, this relies on a player carefully watching those videos and attempting to find patterns and model the behavior (to the extent they can) in the simulator. This is quite time-consuming and prone to errors and biases!

While the generative AI mentioned above is not commercially available (yet), the actual technology has already been demonstrated to be feasible by organizations (like OpenAI and DeepMind) working on such problems. In the medium term, it would advance the game similar to how simulators and web apps (like PokerStars) advanced the game compared to what we had in the 1990s. Ultimately, it would make the game more competitive, and more open to anyone willing to work with these technologies.

Comparing it to robo-advisers in financial investing, can AI be employed effectively to make strategic decisions and generate profits in the gambling sector? What are the challenges and ethical considerations associated with this idea?

For games that involve both luck and skill (such as poker), an AI could be developed in the same vein as a financial robo-adviser to help professional players optimize their earnings. There is a lot of similarity between investing and being a professional poker player.

First, there is significant volatility in both: even the best players are not guaranteed to make money in a tournament, similar to how the best investors can never be completely sure that their calls will yield a profit. Second, poker players have a bankroll on the basis of which they have to choose tournaments and tables, similar to how investors have capital that they need to allocate to yield returns. Both in investing and poker, it’s possible to use leverage, and to buy equity (e.g., in a company or in a poker player). Third, different tournaments and tables have different risk tradeoffs: the World Series of Poker has a large pot, but also higher risk for an average player, since most will never make it to the top 10%. Smaller tables are safer, but with lower returns. Investing is again similar. Investing in a startup is inherently riskier than investing in a safe, publicly traded company.

Given all of these analogies, it makes complete sense for an AI to advise a professional player, based on their stats, bankroll, skill, and previous results against other players in a tournament, on which tournaments to enter and which tables to play at, and for how long. Such an AI could help ordinary (non-superstar) players optimize their bankrolls and actually make a living out of playing poker without spending unnecessary hours ‘grinding’ away, and it could help superstar players. It may also help investors decide who to stake in a tournament to maximize their returns.

I don’t believe there are significant ethical concerns with having such AIs as personal advisers, anymore than there are ethical concerns with having financial robo-advisers. Players and investors should be told clearly that such AIs are advisory only, and there is risk involved regardless.

Related posts