A Strategy for When Humans Must Wage War against AI Computers

We know computers can beat the best human chess players. But what happens when the best algorithms battle each other?

Well, I’ve seen what happens: If both algorithms are equally strong, for example, Google’s AlphaZero versus AlphaZero, the games become wars of attrition. Since neither player will make any critical mistake, the games slowly wind down to a stalemate or a checkmate.

The most important lesson learned from such games, however, is that the player that begins usually wins. The player that makes the first move tends to win the match because of the first-mover advantage such players have.

Why does this matter? It makes machine wars somewhat predictable. If Russia and NATO were to wage war, and if both parties were to rely on equally strong AI to think up their strategy, we know it will turn said conflict into a war of attrition. We also know that the party that struck first shall have the greatest chance of winning.

It matters even more in case human beings find themselves pitted against AI machines. In such a case, there is no possibility of victory for the humans. Humans will make mistakes, and the algorithms may easily erase humanity from history — unless the humans strike first and destroy the machines’ algorithm before it makes a move.

A strategy for humans fighting clever AI machines may be the following:

• strike first (destroy the algorithm)

• if you cannot strike first, buy time by pitting algorithm against algorithm (resulting in a slow war of attrition)

• aim to blow up the “chessboard” (destroy the means of war itself)