It beat humanity’s best Go players and now Google is using football to train next-generation AI technologies

Minghe Hu

After vanquishing the best humanity has to offer in the ancient game of Go, also known as weiqi, Google is now looking to the massively popular game of football to train its next wave of artificial intelligence technology to ‘bend it like Beckham’.

The US internet giant published research in June revealing that its “Brain Team” is working on a game known as Google Research Football Environment to train smart agents that can interact with their environment to solve complex tasks, providing insights into real world AI applications such as autonomous driving and robotics.

Google released a beta version of Football Environment as open-source code on Github earlier this year. The game was built using a publicly available title called Gameplay Football and uses advanced game simulation, including goals, fouls, corners, penalty kicks and offside plays, according to the announcement on Google’s AI blog.

The move comes as technology giants push the boundaries of artificial intelligence technology, a form of machine learning that has been dubbed the fourth industrial revolution, as it moves into more corners of everyday life from autonomous driving, smart city infrastructure and internet of things (IoT) applications to workplace automation.

Why Google’s DeepMind developed Go-playing AI

In 2017, Google outplayed the Middle Kingdom at literally its own game, when AlphaGo, a computer programme from Alphabet's DeepMind Technologies, beat the world’s top Go player Ke Jie 3-0 in a Sputnik-like moment that spurred China into a concerted, state-directed effort to catch up in AI.

Ke ended up in tears, calling the AI programme developed by Google that beat him in three straight matches “perfect, flawless, without any emotions”.

Although the US is generally considered to have the lead in AI, the technology occupies a key role in Beijing’s “Made in China 2025” master plan. According to a three-step road map, Beijing wants to keep pace with leading AI technologies and applications in general by 2020; achieve AI breakthroughs by 2025; and finally to be the world leader in a domestic industry worth US$150 billion by 2030.

The Google AI football research effort is led by Karol Kurach, a machine learning researcher at the US company’s Brain Team in Zurich. Football is considered particularly helpful in terms of AI reinforcement learning, as it requires a natural balance between short-term control and learned concepts, such as accurate passing and high-level strategy, according to Google’s AI blog post.

How China aims to dominate the world of robotics

Google's DeepMind unveiled AlphaGo Zero in 2017, a programme that doesn't need the help of human experts to train itself. Zero is considered stronger than the version that beat Ke Jie because it plays against itself and learns from its own experience.

AI has also been developed by DeepMind and OpenAI to play complex games such as Dota and Starcraft, real-time strategy games for multiple players. OpenAI, an AI research organisation based in San Francisco, California, defeated OG, a world-class e-sports team on Dota 2 using its computer program OpenAI Five, in 2018.

For a comprehensive survey of China’s digital landscape, download the 2019 China Internet Report. Listen to our Inside China Tech podcast and subscribe via iTunes, Spotify or Stitcher. For more insights into China technology, be part of our Inside China Tech group on Facebook.

More from South China Morning Post:

This article It beat humanity’s best Go players and now Google is using football to train next-generation AI technologies first appeared on South China Morning Post

For the latest news from the South China Morning Post download our mobile app. Copyright 2019.