DeepMind, Google’s artificial intelligence (AI) wing, has created a new version of its Go-playing program: AlphaGo Zero. Whereas previous incarnations analysed thousands of human games to hone their play, AlphaGo Zero mastered the Chinese game – in just three days – by playing against itself.
AlphaGo Zero – armed with only the rules – then played 100 games against a previous version that, in 2016, famously beat Go master Lee Sedol. It won them all.
After 40 days’ training, AlphaGo Zero defeated the all-conquering AlphaGo ‘Master’, itself the victor in a showdown with Go’s world number one, Ke Jie.
This represents a major breakthrough because AI no longer has to depend on human knowledge. It could, then, be put to task on some of the world’s most pressing problems.
“For us, AlphaGo wasn’t just about winning the game of Go,” said DeepMind CEO Demis Hassbis. “It was also a big step towards building these general-purpose algorithms.”
“Ultimately we want to harness algorithmic breakthroughs like this to help solve all sorts of pressing real-world problems.
“If similar techniques can be applied to other structured problems, such as protein folding, reducing energy consumption or searching for revolutionary new materials, the resulting breakthroughs have the potential to drive forward human understanding and positively impact all of our lives.”