Photo illustration by Cierra Pedro

AI Could Change More Than the Game(s)

It's a brave new world when it comes to applying artificial intelligence to the gaming and casino industries.

Since the invention of the slot machine over 100 years ago, automation has been a part of gambling, generally for the better. And yet recent developments in AI could substantially shift the Las Vegas resort industry, possibly (though not necessarily) for the better.

Many people like gambling, and a lot of them want to do it as quickly as possible. Slots have evolved from the simple, one-coin-at-a-time mechanical one-armed bandits to electronic, computer-driven devices, but the general function remains the same. Human involvement in slot machines (or electronic gaming devices, which is a more technically accurate term) has, however, decreased. Just 20 years ago, slot machines required an extensive ecosystem of coin and bill processing: Patrons exchanged bills for coins at a cashier booth or with a roving change person, then inserted their coins into machines one at a time. If they were lucky enough to win, they hauled their bucket of coins back to a redemption booth to receive cash.

And there were teams of employees who lugged slot drops from slot bases to hard count, where even more employees sorted, weighed and rolled the coins, starting the process over.

Forgive the tedious description of slot finance circa 1997, but it has a point: Serious slot play was mediated at several points by humans. Players would have to buy coins from a real person, and if their machine ran out of coins or required a hand pay, interact with one or more people.

Now, thanks to bill acceptors and ticket-in/ticket-out technology, players can go hours without needing human intervention. Bills are put directly into the machine, which spits out a ticket instead of coins. The only time a person is necessary is for a high-value hand pay or a win that triggers the IRS’ reporting requirements.

That evolution has changed casinos tremendously. There are many positives for the player, including more convenience and quicker play. If you want to know about the positives for the employees, just ask around hard count, if you can find anyone.

The self-driving Arma shuttle bus debuted on Fremont Street for a 10-day test run in early January 2017.

It’s no secret that more and more tasks are becoming automated. French startup Navya’s self-driving shuttle is making its rounds Downtown. Check-in kiosks in Strip lobbies are making front-desk attendants a little less critical. Not too far off are machines that pour drinks, prepare meals and clean rooms.

Historian Hal Rothman believed that Las Vegas was “the Last Detroit,” the final remaining place in postindustrial America where hardworking but low-skilled men and women could pull down a middle-class wage. As casinos opened in the 1990s, the expanding subdivisions of a growing city were filled with valets, dealers, housekeepers—everyone needed to keep the machine running.

But, if a robot surgeon can outperform a highly trained human doctor, is it really that hard to imagine that it won’t be able to mix a cocktail better, too? And, with no sick days or free pours. Just as with free parking, once one company successfully rolls the dice on what promises to be a labor-saving innovation, the others will almost surely follow suit.

That means no more “Last Detroit,” which will fundamentally change the nature of casino employment. It is strange (but totally implausible) to imagine a future in which more professional hockey players than bartenders work on the Las Vegas Strip—something completely in the realm of fantasy only a few years ago.

New developments promise (or threaten) to extend automation further up the organizational chart. Recently, Google DeepMind’s artificial intelligence AlphaZero beat Stockfish, the world’s best computer chess engine, 64-36 across 100 games. Stockfish could fight AlphaZero to a draw about of third of the time, but it couldn’t beat the AI.

One computer beating another at chess might not seem like a big deal, but this was a significant contest. Stockfish is the reigning computer chess champion, meticulously programmed to respond to a variety of sequences. AlphaZero, on the other hand, was not taught how to play chess at all. Supplied with the game’s rules, AlphaZero, using machine learning, taught itself to play chess by playing against itself. Four hours later, it was ready to dethrone a champion. Most impressively, it used “humanlike moves and ideas” to win.

AlphaZero’s success means that people who get paid to make decisions might be no more immune to automation than those who work with their hands. If artificial intelligence can’t be designed to assess risk or manage resources better than humans, it now appears that it can teach itself to do it—in the time it takes to drive to Las Vegas from Los Angeles.

That means much, much more than the possible loss of the bulk of executive jobs in Las Vegas resorts. Because if there are fewer managers working worldwide, that means there are fewer managers with money to vacation in Las Vegas.

Which is all a convoluted way of arriving at the conclusion that automation is going to severely disrupt the global economy in ways that will help some but not others. This will spark greater social and economic shifts that could benefit or hurt Las Vegas. Perhaps all of the blue, white and pink-collar workers displaced by automatons and AI will find new work that pays better, giving them more time and money to spend in Las Vegas.

Or maybe not. Either way, Las Vegas tourism will have to adapt to the new realities. It won’t be the first time: The industry took advantage of many national trends to grow in the second half of the 20th century, and skillfully adapted to the proliferation of casino gambling to remain relevant in the first years of the 21st.

It might take longer than it would an AI, but Las Vegas will hopefully learn how to survive.