Chapter 460 (1/2)

At the beginning of AI stepping into the field of games, it didn't go well.

Before the company developed origin, it also developed another artificial intelligence to step into the RTS field, which was completely hammered by human players.

Of course, the main reason for choosing RTS as the main direction of attack is the type of game.

At present, the only games on the market are RTS games, in which human beings can fight against AI fairness.

Like FPS games, no matter how hard we try, we can't play the plug-in game; and leisure puzzle games, we can't win anything.

Those seemingly very difficult puzzles are just very difficult for human beings, and completely no difficulty for artificial intelligence.

After all, there are so many educational games in the world, which one can surpass the difficulty of go?

As for the now popular MoBa game, it's a 5v5 game after all. Let five human players play five artificial intelligence with exactly the same idea? It doesn't feel right either.

So, the company that studies artificial intelligence chooses to choose, only RTS game is most suitable.

The key element of RTS is the fog of war.

In RTS, due to the limited vision, artificial intelligence can not grasp all the movements of the opponent, which gives both sides space for strategic game.

At the same time, because there are extremely complex attributes among various arms, many times this change is faster and more complex than go. If you don't find or understand a small intention of the other party in time, you may be pushed flat by a wave of direct dragging.

For AI, the biggest limitation is always the speed of operation.

At the beginning of the battle between AI and go, human beings can surpass AI when playing fast chess, but once playing slow chess, they will be defeated. Because artificial intelligence and human way of thinking are totally different, the more complex the situation, the longer it takes to think before calculating the optimal solution.

In real-time strategy games, there is not enough time for thinking like go. For AI, there may be another completely different algorithm and rule.

Of course, in addition to these aspects, in terms of operation, artificial intelligence obviously completely crushed the human players.

For some human players, APM in RTS games is never enough, because dozens of units need to be operated at the same time, even if the brain is enough, the hand can not keep up.

But for artificial intelligence, no matter how many units are operated, and no matter in terms of reaction or accuracy of operation, all of them are human competitors.

Therefore, considering that this is relatively unfair for human athletes, the company that developed origin also made certain restrictions on the operation of artificial intelligence, and forced its APM average value to be limited to about 400, and the peak value to be limited to 600, which is the same as the first-class human experts.

Of course, for AI, even if APM is limited, it has an advantage over human beings, because it does not need to make those repeated operations to warm up the hands, and every step of its operation is an effective operation.

In other words, the average APM of origin seems to be about 400, which should be equivalent to 500 or even 600 of human beings.

After origin successfully climbed the ladder of three servers in the United States, Europe and South Korea, origin's R & D company announced that, considering that origin has too many advantages in APM in the face of human competitors, it is considered to continue to reduce origin's APM value to about 350 on average.

Of course, this can be said to be for the sake of fairness, or it can be said that the R & D company of origin is afraid to scare those professional players away, no one will take up the challenge, or fear that the man-machine battle will become one-sided and unappealing, so it makes such adjustment.