- DeepMind announces the launch of a new AI program, MuZero
- MuZero successfully learns the rules of a game as it plays
- The new AI program is currently being tested with YouTube
What is DeepMind?
DeepMind Technologies is an artificial intelligence company and research laboratory founded in September 2010. The London-based startup, which was acquired by Google’s parent company, Alphabet in 2014 for $500 million.
Since then, DeepMind has reached some remarkable achievements in the field of artificial intelligence. The company has been hailed by leading AI experts for pioneering the development of human-level artificial intelligence.
DeepMind’s New AI System – MuZero
In a recently published paper titled “Mastering Atari, Go, chess and shogi by planning with a learned model” DeepMind detailed how their new AI system operates. DeepMind first published details of MuZero in 2019. It was never explained until the release of the paper in the article journal, Nature.
Sponsored
MuZero, DeepMind’s new AI program was shown to be capable of learning the rules of a game (chess and shogi) as it plays. To achieve this feat, MuZero utilizes a principle called “look-ahead search”.
When considering future moves, the prioritize, narrowing moves down to just the most likely and relevant moves. Instead of modeling all the possible moves, MuZero learns from successes and failures then narrows down its decisions to the most relevant.
On the Flipside
- In MuZero, DeepMind has created a new benchmark for what is possible with artificial intelligence. Although its new program is record-breaking, the path had been previously charted by;
- DQN, a program that achieved human-beating proficiency in Atari video games using only pixels and game scores as input
- PTC’s (Parametric Technology Corporation) latest AI software helped NASA design a new moon-bound space suit.
David Silver, the principal research scientist at DeepMind, believes that MuZero is the first AI mode capable of generating its representation of rules of an environment. Using those rules, it can plan out future actions. In an interview with TechXplore, he explained:
For the first time, we actually have a system that is able to build its own understanding of how the world works and use that understanding to do this kind of sophisticated look-ahead planning that you’ve previously seen for games like chess.
According to Wendy Hall, professor of Computer Science at the University of Southampton, MuZero represents “a significant step forward” for reinforcement learning algorithms.
How Will DeepMind’s latest AI Program Enhance YouTube
In an interview with BBC, DeepMind’s principal research scientist, David Silver announced that their new AI program is currently being tested by YouTube.
According to David Silver, tests are ongoing to find a new method of encoding videos on YouTube. If the AI prowess of DeepMind is utilized by YouTube, we will see faster encoding times. In turn, that could slash the cost of YouTube services.
