Google and DeepMind lab’s artificially intelligent AlphaGo program has become a little more intelligent on its own. The program has achieved many-a-feats such as beating human players at one of the most complex board games known Go. Now, as the program matures into AlphaGoZero, the program has taught itself the game with no human input. Also Read - New avatar of Google Chrome’s offline dinosaur game: How you can playAlso Read - Best camera phones under Rs 35000 to buy in July 2021: Pixel 4a, Mi 11X, and more
Go is a traditional Chinese board game that has players implement complex strategies to surround more territory using colored stones than the opponent. AlphaGoZero, which is the new version of AlphaGo, has improved itself with deep reinforcement learning to beat human players at the game. A DeepMind blog post says how the system starts with a neural network which does not know anything about Go game. It then plays against itself, combining the neural network with a search algorithm. When combined, a new version of AlphaGoZero is produced. This process is repeated to build a better program at the end of each iteration. Also Read - Timex Helix Smart 2.0 with temperature sensor, heart rate sensor launched: Details here
Over a period of a few days, the program played millions of games against itself and learnt the Go game. In the process, not only the program taught itself the game from scratch, but also learnt human strategies and a new type of knowledge which were unconventional for humans.
AlphaGoZero’s competence was recorded at a superhuman level when after 70 hours of training, it could play the complex board game involving multiple challenges across the board. ALSO READ: Facebook s AI bot CherryPi loses in StarCraft: Brood War competition, but shows impressive skills
It is not the first time that the AI program has shown exceptional skills in playing the complex game. Earlier this year, AlphaGo beat the world’s number one player of the ancient Chinese board game Go, signifying a major breakthrough in artificial intelligence.
AlphaGoZero or any such variations of the core AlphaGo program are not just aimed at beating humans at board games. The potential of the program lies beyond gaming strategy and into solving some real-world problems. Demis Hassabis, co-founder and CEO, DeepMind, says,”It s amazing to see just how far AlphaGo has come in only two years. AlphaGo Zero is now the strongest version of our programme and shows how much progress we can make even with less computing power and zero use of human data. Ultimately we want to harness algorithmic breakthroughs like this to help solve all sorts of pressing real-world problems like protein folding or designing new materials. If we can make the same progress on these problems that we have with AlphaGo, it has the potential to drive forward human understanding and positively impact all of our lives.” ALSO READ: One in three smartphones to be AI-ready in 2020: Counterpoint Research