[ad_1]
The AI agents started making random moves, but after playing again and again, they finally figured out the strategies and techniques that work. Wojciech Czarnecki, researcher at DeepMind, said NYT that they can even "adapt to their teammates with arbitrary skills". The result? According to the newspaper, the agents managed to compete with human players even when their reaction times were slowed down compared to their opponents of flesh and blood. They "clearly exceeded the rate of victory of humans on cards that neither the agent nor the man had seen before," reads the newspaper. This is a huge win for DeepMind because it is much harder to train an AI for tasks requiring people to work in groups. And it's a problem for the company to find a solution to be able to develop automated systems for the real world.
As The New York Times noted that DeepMind's skills that AI is recovering from complex games like StarCraft II could be used by warehouse robots that have to work together. They could also lead to a solution that would allow autonomous cars to navigate en masse in heavy traffic. DeepMind still has a lot of work ahead of him. Georgia Riedl, a professor at the Georgia Tech College of Computing, told the NYT The problem of the artificial intelligence agents of the company, is that they only react to everything that happens in the game. They do not communicate with each other as do the groups d & # 39; # 39; humans and animals. That is why we would not be surprised that the company is adopting in the future more multiplayer and complex video games, with the aim of strengthening its artificial intelligence capabilities.
[ad_2]
Source link