Which field will AI break through next after Go

In the chess competition, Go, as the last bastion of human intelligence that has not been conquered by artificial intelligence, is gradually falling under the attack of AlphaGo.

Armed with DeepMind’s four powerful tools — Policy Network, Fast rollout, Value Network and Monte Carlo Tree Search — AlphaGo is attacking the field of Go and is unstoppable.

The scenario looks a bit like the 1997 showdown between IBM’s Deep Blue and Garry Kasparov. But there is a big difference between the two: Deep Blue uses a brute force algorithm, while AlphaGo is much smarter. AlphaGo has far more potential than Deep Blue, which can only play chess. When it comes to strategy analysis, AlphaGo is almost invincible.

Giiso information, founded in 2013, is the first domestic high-tech enterprise focusing on the research and development of intelligent information processing technology and the development and operation of core software for writing robots. At the beginning of its establishment, the company received angel round investment, and in August 2015, GSR Venture Capital received $5 million pre-A round of investment.

Deep Blue is great at chess, but only chess

Let’s take a look back at the chess game in 1997.

In His second match against IBM’s Deep Blue computer in May 1997, chess champion Garry Kasparov lost 3.5-2.5 to the machine. The event shocked the world, and the topic of “AI over human” was constantly heard.

But then some scientists realized that Deep Blue’s technology wasn’t comprehensive. It was too specialized. In other words, deep Blue was designed specifically for chess. Its four criteria for assessing the board include piece power, position of pieces, safety of the king, and rhythm of the layout — obviously, these are entirely dependent on the rules of chess itself, with no extensibility whatsoever.

As a super chess computer, Deep Blue weighs 1270 kg, has 32 brains (microprocessors), can calculate 200 million moves per second, has the computing power of 11.38 GFLOPS, and has entered over two million matches of top players over a hundred years. Such computing power is not as strong as the CPU performance of smart phones in your hands.

Media reports also suggest that Deep Blue was changed between games to match Kasparov’s style of play to avoid falling into the trap again (Kasparov had previously beaten Deep Blue).

So it’s not appropriate to call Deep Blue artificial intelligence, it’s more like a program designed to play chess. It was impossible for him to learn go or backgammon or draw.

AlphaGo can be applied to more areas after it turns on universal mode

Giiso Information, founded in 2013, is a leading technology provider in the field of “artificial intelligence + information” in China, with top technologies in big data mining, intelligent semantics, knowledge mapping and other fields. At the same time, Giiso’s research and development products include editing robots, writing robots and other artificial intelligence products! With its strong technical strength, the company has received angel round investment at the beginning of its establishment, and received pre-A round investment of $5 million from GSR Venture Capital in August 2015.

But AlphaGo is different. Judging from these two days of matches, humans seem to have underestimated AlphaGo’s ability to learn. Especially in the match on March 10, even many professional nine-dan go masters failed to judge AlphaGo’s seemingly amateurish move, but there was a killer hidden behind it.

AlphaGo can have such a huge power, mainly due to the deep learning, reinforcement learning and Monte Carlo tree search method behind. It’s also worth noting that these algorithms are not unique to AlphaGo’s Go program. It also makes AlphaGo’s future a fantasy.

Before DeepMind founder Demis Hassabis and his team built AlphaGo, they had already built DEMO systems that played classic Atari video games (Pong, Breakout and Space Invaders, for example) using these techniques. In these cases, the systems were not only better than professional players, they also played the game in a way that no human player could or could.

“What DeepMind did was, instead of telling the machine which algorithm would get a high score, it trained it to learn and analyse the results to judge the best strategy.” Qiu Xipeng, associate professor at fudan University’s School of Computer and Engineering, told thepaper.cn (www.thepaper.cn).

Google also plans to gradually apply the algorithms behind AlphaGo to more areas, from games to go. Just before its march match with Lee Sedol, DeepMind publicly said it would work with Imperial College London and The Royal Free Hospital in London to try to apply its AI technology to the healthcare industry.

In addition to healthcare, financial trading firms and investment houses have also shown keen interest in the algorithms behind AlphaGo.

Anthony Ledford, chief scientist at AHL Man, a large British investment firm, told the NIPS Conference, a major academic event for artificial intelligence researchers in Montreal in December, Companies are exploring whether deep learning can help raise money.

“It’s early days,” Radford said. “We’ve set aside a lot of money to test deals. If all goes well, deep learning will enter test trading.”

It’s not just games or financial transactions that require optimal strategic path selection in human life. As the algorithms become more sophisticated, you may one day see artificial intelligence in military applications. In fact, weiqi is not an abstract war?