AI – AlphaGo Go

Much emphasis and interest have focused recently on where an AI source can act as the inventor instead of a natural person. The assumption is that an AI source can invent, that is to say, do something which isn’t obvious and so has an inventive step. These arguments can be a little esoteric in nature.

Care needs to be taken that merely quicker and faster isn’t mistaken for intelligence. An example of this is the entity known as AlphaGo.  AlphaGo is stated as providing artificial intelligence in relation to playing the game GO which is a strategy game apparently far more difficult than chess. The AlphaGO entity considers the moves made in a vast number of previous grandmaster games and so from the current position provides its servant (the human who actually moves the pieces) with the move which will in all likelihood lead to a winning position.

It is intelligent in the sense that it responds to the opponent’s next move, but again it is based upon what is best from prior grandmaster games and will produce a win. It is believed that there is further sophistication such as considering several of the opponent’s previous moves.

Essentially, the AlphaGO entity uses faster and greater computer capacity for a trial and see approach.  The equivalent would be a human with infinite memory capacity to remember all previous GO games and so, given time, that human applying no emotion just plain statistics, could choose the next move. That is not inventive in itself but the next stage is to introduce emotion, a hunch you might say so that it’s not just statistics but hopes to get to a win quicker or in fewer moves despite a greater risk of losing – that might be inventive.