Super-AI gaming inspires people to develop innovative, effective tactics

Share This Post

This is probably also the first time that humans have to face the fact that an inanimate object can be smarter than them.

It all started at the New York World’s Fair in 1940, when Edward Condon, a nuclear physicist, presented a one-ton electromechanical machine that successfully played a simple ancient Chinese game called is Nim, in which two opponents remove objects from the stack. until only one remains. The machine, one of the first computer games ever recorded, easily transports with most of its human challengers.

Fast-forward six decades, and IBM’s Deep Blue coin-sized computer conquered world chess champion Gary Kasparov, winning three games and drawing one. Ten years later, the tournament challengers were defeated by an AI chess program hosted on a mobile phone.

Artificial intelligence developed rapidly in the 2000s, resulting in massive computer systems capable of learning to play on their own by analyzing popular and complex video games.

For example, computerized versions of Go – a seemingly simple game, but many times more complex than chess – have for many years been best played only at the amateur level. But DeepMind’s AlphaGo has used neural networks that bring a new dimension to AI, analyzing 10,170 possible card layouts to create a decisive program that wins over the top masters of the game. play.

But according to a research paper published in the Proceedings of the National Academy of Sciences, humans can win, or even beat, superintelligent programs. Minkyu Shin, assistant professor of marketing at City University of Hong Kong, said the advent of superhuman artificial intelligence has pushed humans to become more creative in game strategies and is responsible for improving improve their game play.

Shin and his fellow researchers conducted a large analysis, sifting through a database of 5.8 million Go moves made over 71 years, starting in 1950. (when the rules of Go were standardized).

“Our results suggest that the development of superhuman AI programs may have pushed players away from traditional strategies and motivated them to explore new moves, thereby increasing the odds of a game being played,” said Shin. can improve their decision-making.” They developed a formula to rank the quality of human decisions at each stage of the game. Using KataGo, a superhuman AI program, they compared human winning moves against nearly 60 billion theoretical game samples. This produces a score known as the Decision Quality Index (DQI).

By examining DQI scores over decades, Shin and his team found that even though human players made minimal progress in the first decades of Go, significant improvement was achieved. seen shortly after 2016, the year of AlphaGo’s remarkable early achievements. “We found that human decision-making has improved significantly after the advent of superhuman AI,” said Shin. “This improvement is related to more novelty in human decisions.”

Related Posts

Concerns About KFC’s Diablo 4 Beta Codes Have People Panicking

While some users are completely against it, others are...

Microsoft’s new Copilot will fundamentally alter Office documents

I was talking to Friedman on a Teams call...

Microsoft offers EU remedies to get the deal with Activision approved

According to Microsoft President Brad Smith, the American software...