I’m no brain scientist but it tells me that computers/machines continue to improve in their ability to compute exactly as one would expect from modern computers doing modern computing. The algorithms used and the math and statistics which comprise them have gotten more complex, clever, and sophisticated, as have the programming languages and code from which those languages are built. However, at no point do algorithms and math and statistics and programming and code coalesce or emerge by some magical but never defined process into intelligence. It is still a machine, and it is a computing machine, it is a computer, no more, no less, and it is not intelligent in any sense of the word.
No matter how sophisticated the game there is one thing that all games have in common, they have rules, as Wittgenstein so ably showed us. Computers are rule following machines, they are experts at following rules. In fact one could argue they are the optimal form of a rule following “thing.” i.e. nothing could be conceived of or built that would be better at following rules, it is in a computers “DNA’ to use a terrible and wrong analogy to follow rules because that is all they can do, can ever do. It is no surprise that computers continue to get better at games because they continue to get better at following rules. If a computer were ever built that was able to stray from the rules given it in its programming it would no longer be a machine. At that point we would actually have AI and a new form of “life”. We are a very, very long ways from such a day however, if it is even possible.
Cognitive neuroscience already fell victim to the mereological fallacy and now from that grave mistake another logical error has emerged, the compulogical fallacy. How many logical fallacies can any one field of study embrace?