The latest encounter between man and machine took place in Hainan, China, and involved a total of 36,000 hands. That was considerably less than the 120,000 hands played by Libratus in January, but the poker in China was an exhibition, not a scientific experiment. Even so, the resulting numbers showed that Lengpudashi emerged from the series with a bigger margin of victory, as measured by the numbers of milli-big-blinds (the standard metric used to compare poker efficiency) won per game, which is definitely encouraging for AI developers everywhere.
This success confirms that artificial intelligence is not only capable of beating professional poker players, but that it is also improving. Back in 2015, an earlier Carnegie Mellon program called Claudico took on four professionals and lost, so many were doubtful that its successor, Libratus, would fare any better.
Of course, Libratus fared much better, soundly beating its human opposition and making history at the same time. Now Lengpudashi has also beaten its human opponents, and we highly doubt that the evolution of artificial intelligence will end there. On the contrary, we can more probably expect AI to continue to improve at an exponential rate.
The $290,000 in prize money which was won by Lengpudashi is now in the hands of Strategic Machine Inc., the company which was formed by Carnegie Mellon computer science professor Tuomas Sandholm and Noam Brown, who is a PhD student. It was those two individuals who actually developed the AI systems which have proven to be so effective, and the money will no doubt help the company to continue doing more of the same.