Name: Artificial intelligence, solvers, clankers, ‘slop’.
Age: 73.
Appearance: The toughest opponent you’ll ever face, or the best coach you’ll ever have.
Wait, it’s 73 years old? More or less; the first computer that could play a game of checkers was created in 1952 by Arthur Samuel. The first publications discussing the concept, including Alan Turing’s groundbreaking Computing Machinery and Intelligence, emerged a few years before that, and the term ‘artificial intelligence' was coined several years after, but when it comes to AI in games it really began with the ‘Samuel Checkers-Playing Program’.
Catchy name? Was it any good? It was — in 1961 the program took on one of the best checkers players in the USA, and won.
But how did it pick up the pieces? Don’t be obtuse. The program was loaded with reams of information from past games to determine the best moves available at any specific point, the pieces were moved by mere humans (it seems AI has always had an issue with fingers). Samuel’s work went on to be of huge importance to his employers, a little computing firm known as IBM.
IBM was around in the 1960s? Indeed. In fact, it has been around since 1911, when it went by the name of the Computing-Tabulating-Recording Company. It would change its name in 1924 to the International Business Machines Corporation, then IBM for short. The company would go on to employ another key figure in gaming, with Isai Scheinberg working there as a programmer before leaving to co-found PokerStars.
So did the careers of a generation of checkers players go up in smoke? I don’t think the pro checkers circuit is in great shape, but I’m not sure you can lay the blame at AI’s doorstep.
How about poker? In terms of strategic depth, comparing poker to checkers is like the Atlantic Ocean and that puddle of rainwater in your garage that never seems to go away. Poker took a lot longer for computers to get to grips with, although the last 10 years have seen progress eventually overtake human capabilities. Bots with names like Cepheus, Claudico, Libratus and Pluribus have continually pushed the boundaries when it comes to out-performing human players.
Why do they all sound like Roman Emperors? It seems developers are fans of latin. For example, ‘Claudico’ is latin for ‘I limp’, while ‘Libratus’ means ‘balanced’.
Clever. You can say that again; Pluribus consistently beat a number of poker pros in a multiway game in 2019, truly ushering in the era of ‘superhuman’ poker.
What does this mean? Is it a good thing, or a bad thing? That depends on how you look at it.
*Sigh*... I thought you’d say that. It’s true. AI-powered bots are clearly not opponents you want to mess with, but that’s why online sites invest in security measures to combat them, using both AI and human poker experts to flag and investigate suspicious play and winrates. What you should be thinking about is how AI can help you become a better poker player.
Isn’t that against the rules? Using real-time assistance is absolutely against the rules, and will see you banned from the site — and deservedly so. But using non-real-time assistance, i.e. when you’re not playing, is heartily recommended. Things like strategy books, podcasts, 1-to-1 coaching sessions and, yes, AI can be valuable ways to spend your study time.
Okay, but how? I’m not exactly the most tech-literate person in the room (even when I’m the only one in it). There are plenty of options these days. Software like GTO Wizard and Octopi Poker help you really drill down into complex concepts and tricky spots for a deeper understanding. If you’re interested in playing game-theory-optimal (GTO) poker, which you should be, they’re well worth investigating.
Is this taking away the ‘creativity’ of poker? Did Doyle Brunson do that when he published Super/System and opened the door to a new generation of poker study? Like a good book, AI can help you see things in different ways and expand your horizons as a player. Consider the words of Lee Sedol, the high-level Go player who lost to an AI in the 2016 documentary AlphaGo, after it made a move he would never have even considered: “I thought [the AI] was based on probability calculation, and that it was merely a machine. But when I saw this move, I changed my mind... This move was really creative and beautiful.”
Do say: “I think you'll find, according to the solver…”
Don’t say: “...I should give up poker and go play checkers.”
