New AI Can Win Multiple Games With No Prior Instructions
MuZero can master games like chess without instructions. Credit: DeepMind
As days pass, new technology is created that ether teaches us something new, extends what we know now, or literally doesn’t don’t do anything. But in the artificial intelligent category, a company named DeepMind has created a new AI that will rock the scene.
On December/23/2020, DeepMind made their latest AI, MuZero. MuZero is the predecessor of AlphaZero which had won against Fan Hoi, the raining champion of the ancient Chinese game of Go. Although AlphaZero was a great achievement, the AI had a problem, AlphaZero could only play new games if it had pre-programmed knowledge of how the game work.
How MuZero Works
So this is where MuZero program steps in. MuZero will try one action after another until it learns the rules of the game. While learning the rules of the game though, MuZero will start noticing the rewards and penalties of each action it does. MuZero will then start to refine its strategy until it hits the jackpot finding the winning approach. If you want a more complex explanation please look at the video above.
DeepMind has already used MuZero in inventing better video compression, and will start to let MuZero do more real world applications like protein design and automated vehicles!
“The pace of progress in artificial intelligence is incredibly fast. Unless you have direct exposure to groups like Deepmind, you have no idea how fast—it is growing at a pace close to exponential. The risk of something seriously dangerous happening is in the five-year time frame. 10 years at most.”
—Elon Musk
