It's not often that board games make it to the national news but this week, most of you probably saw the big announcement that a Google AI lab has developed a program that has beaten the European Go (also know as Wei Qi) champion.
This is indeed an astonishing achievement. Go has only 3 basic rules so is beautifully simple and aesthetic but is more complex than Chess - the number of permutations and options quickly eclipses those of Chess by orders of magnitude. When Deep Blue 2 beat Gary Kasparov (who is generally accepted as the best player ever to have lived), it was pretty much a brute force approach - the computer was simply bashing out every possible permutation as far into the future as possible.
You can't do this with Go - the permutations quickly spiral exponentially out of control. I did a quick calculation; for the first 3 moves there are almost 50 million possible outcomes! So the achievement here is considerably more special than IBM's. This program actually learns. Those of a paranoid nature may be concerned because slightly worryingly, although the initial algorithm analyses real-life games played by actual humans, (fleshy beings of little value), thereafter, it learns by playing itself... Those who haven't thought about this before may at this point wish to rush off and watch some Sci Fi - I would recommend 'War Games' from the 1980s, the 'Terminator' series or the excellent 'Ex Machina'.
Fortunately for us fleshy creatures, I recall from my computability theory course that it has been proven mathematically that humans think in a different way to computers. There are some problems that humans easily manage that it is impossible to write a program to solve.
Back on earth, those of us who are interested in these things are looking forward to the match between Google's AI and the current Go No. 1, Lee Sedol in March. There isn't any other game that is considered to be more difficult than Go, so if the computer wins, a landmark will have truly been reached.