Artificial intelligence is regarded as one of the defining technologies of the future. Depending on who you ask, it’s a technology with tremendous potential for the benefit of human society. But others foresee a darker future, almost sci-fi dystopian in its outlook, that envisages artificial intelligence growing to rival human life.
In reality, AI will probably bring with it both the good and the bad, like so many modern technological innovations. But there is one group who might want to reflect on some recent news and what it might mean for them – poker professionals. An AI programme has successfully demolished all comers in a series of poker contests in China, taking on six of the region’s most successful poker players in the process.
Through deploying its own strategy, the programme was able to outfox its competitors, including some players who have previously won in the World Series of Poker. When the dust settled, the programme had amassed winnings of $290,000 across five days of action, showing the sophistication of the technology behind the results.
Lengpudashi’s Incredible Poker Brain
The AI programme in question was called Lengpudashi, the second example of a competition-busting poker AI to be trialled this year, following the success of its predecessor, Libratus, earlier in the year.
It has been developed by computer science profession Tuomas Sandholm, in conjunction with PhD student Noam Brown at Carnegie Mellon University in Pennsylvania. The pair have created a company to take forward their creations, Strategic Machine, which will be the recipient of the prize money their programme won over the series of exhibition matches.
Lengpudashi was pitted against some stiff competition from the world of competitive poker, including Yue Du, who won the WSOP $5k no-limit Texas Hold’em category at the 2016 event.
In addition to competitive poker talent, Du’s ‘Team Dragon’ also drew on expertise from computer scientists, engineers and AI experts, to attempt to second guess the AI, in a concerted attempt to beat the script.
The AI’s victory has been celebrated in technical circles, and gives a strong signal as to the sophistication of the system behind its decision making. Of particular note was Lengpudashi’s ability to bluff weak hands successfully, having learned that it is possible to maximise its winnings from poorer quality hands by deploying successful bluff techniques.
The news also speaks to the wider development of artificial intelligence, which some analysts believe could come to replace significant swathes of exclusively-human tasks, leading to something of a new technological revolution in the decades to come.
Imperfect Information Games
AI has previously successfully defeated players in games like chess and Go. But there is a key distinction with poker, in that many of the variables remain hidden from view – unlike these examples, poker is a game of imperfect information, which requires a level of AI sophistication that goes far beyond.
In addition to learning how to bluff of its own accord, the AI was also required to read its opponents moves, and to make assessments about when opponents might have been bluffing, while computing multi-layered betting decisions. While no doubt the team at Strategic Machine will be looking to continue to refine and develop their technology, these performances will do wonders for confirming the capacity of AI to give even crack-teams of humans a run for their money.
An AI Future?
Does this mean AI will be able to beat other poker variants such as PLO, 8-Game or even classic video poker games and if so, should poker professionals start to worry? Well yes, at least if they hope to challenge AI competitively in the near future. While there will remain academic interest in giving AI as rigorous a challenge as possible, it’s unlikely that your average poker player will find themselves losing out just yet.
Unfortunately, it’s highly possible that AI this sophisticated could take on a range of functions in the real world. Even highly specialised roles, like lawyers and accountants, could ultimately see much of the demand for their services replaced by AI, while manufacturing processes and other, arguably more basic functions are already witnessing the AI revolution first hand.
Self-driving cars are likely to spell huge fundamental change to various transport industries, while construction and infrastructure development look set to become significantly more efficient in future. While AI technology will no doubt prove a powerful tool in the years and decades to come, its capacity for radically changing society cannot be underestimated.
And with incredible poker brains like Lengpudashi already making waves, it’s impossible to imagine how successful AI machines could become in the near future, as technology continues to improve.