Back in 1997, a chess-playing computer called Deep Blue faced off against world champion Garry Kasparov in a series of six games, eventually winning the head-to-head duel 3.5 to 2.5 after one of the matches ended in a stalemate.
Now, a similar artificial intelligence (AI) programme is being set up to pose opposition to humans in a popular casino game.
Almost a year ago to the day, we discussed the concept of robot dealers being introduced to casinos, alongside the pros and cons of such a model.
After all, they would bring a novelty value and intrigue that would likely increase visitor numbers, at least in the short term. Robot dealers would also be more efficient and cost effective, although the lack of human interaction would undoubtedly remove some of the enjoyment from games.
The discussion now moves to the opposite side of the table.
Blackjack is the game that naturally lends itself best to playing like a robot. After all, the most successful players look to negate the house edge working in the casino’s favour by following a rigid strategy based on probability.
Based on the dealer’s upcard, players will know instantly whether they should hit or stand on any given hand, whether they should be splitting their pairs or not etc.
For example, if a player is dealt two aces they should always be split, if dealt two cards with a face value of 10 they should never be split or if dealt two nines it is advisable to split against any dealer upcard which is not a seven, 10 or ace.
It may take ample hours to memorise all of the outcomes within what is known as “Basic Strategy”, but the suggested plays are all based on computer simulation testing and are considered to limit the losses expected because of the house edge.
So this is why it isn’t disadvantageous.
The latest AI test to be conducted focused solely on poker and has the potential to mostly impact upon the online casino experience.
At a casino in Pittsburgh, Pennsylvania, four professional poker players tackled 120,000 hands between them spread over 20 days against a robot called Libratus.
However, setting up the algorithms presented a considerable challenge for programmers because of the amount of actual information available at a given time at poker.
At chess, there is no hidden information which was needed to be built into an algorithm.
But poker is a game where the art of deception and bluffing is a substantial factor, making it hard to pre-programme the computer to react to unknowns.
Yet by the end of the gaming marathon, the computer was miles ahead, up 1.7 million chips. This was the sum of 17,000 big blinds. Each of the four poker pros found themselves in a losing position at the end of their stint of heads up no limit Texas Holdem.
The good news for online casino players is that there is no plan to ever use Libratus outside of testing the power of AI and the impact it could have on other real world situations, but the obvious problem is that others could attempt to replicate the technology used.
Also, so far the algorithm has only been tested during heads-up poker play and the majority of online games feature tables containing many more players. Sometimes this can be as many as 10.
It is believed that the robot could manage fine in a three-player game, but even a six-player setup would be beyond it.
However, Noam Brown, one of those involved in the development of Libratus, told Card Player: “The annual computer poker competition is adding a six-player league going forward, so research on six-max poker is going to start to happen and I think that the field is going to develop very quickly.
“I think that with some minor improvements to Libratus, you’d be able to see it beating humans at six-max within two years.”
With two players, the focus tends to be optimising game theory, whereas the introduction of more players could see the main strategy adapted to focus primarily on something else, such as milking the most from weaker individuals at the table.