Artificial superintelligence will help solve humankind’s intractable problems and it’s being built as we speak.

Oh no, I hear you say, not another success at yet another game by DeepMind or Meta or whichever. Haven’t we seen all this before with AlphaGo and Pluribus? Well, yes and no. You have got to start somewhere. Like acorns and oak trees, you start with small games and then you scale it up. The latest big, big games to be toppled by AI are Stratego and Diplomacy. The point about these games is that the number of combinations of moves puts playing Go and poker in nursery school. AND numerous other players are playing at the same time. Yet AI still mastered them. Rather than using search trees, it has been designed to use other techniques such as reinforcement learning based on actual outcomes from repeated game play. So, the daunting number of combinations of possible moves becomes irrelevant. More importantly, these breakthroughs demonstrate that AI can tackle big problems involving many players such as, guess what, climate change negotiations at a COP near you. This will be one of the leading benefits of artificial superintelligence when it emerges.

Link to article:

You may also like to browse other AI posts: