据《金融时报》2月18日报道,业余围棋选手凯林·佩里恩(Kellin Pelrine)利用另一台计算机发现的先前未知的AI缺陷击败了AI选手。在这场对弈中,佩里恩赢得了15局中的14局,且是在没有计算机的直接支持下进行的比赛。
A powerful amateur Go player has defeated a high-level AI system after exploiting a weakness discovered by a second computer.
一个强大的业余围棋玩家在利用了第二台电脑发现的弱点后,打败了一个高水平的人工智能系统。
Kellin Pelrine, a player who is one level below the top amateur ranking, beat the machine by taking advantage of a previously unknown vulnerability that had been identified by another computer. But the head-to-head battle in which he won 14 of 15 games was made without direct computer support. It’s a rare Go win for humans since AlphaGo’s win in 2016, which helped lay a solid foundation for the current AI popularity.
Kellin Pelrine是一名低于顶级业余排名一级的球员,他利用另一台计算机识别出的先前未知的漏洞击败了该机器。但是他在15场比赛中赢得了14场的正面交锋是在没有电脑直接支持的情况下进行的。自2016年AlphaGo获胜以来,这对人类来说是罕见的胜利,这为当前AI的普及奠定了坚实的基础。
Pelrine’s victory was made possible by a research company called FAR AI, which developed a program to check KataGo for weaknesses. After playing over a million games, it was able to find a weakness that could be exploited by a decent amateur player. Pelrine used the same method to beat Leela Zero, another top Go AI.
Pelrine的胜利是由一家名为FAR AI的研究公司实现的,该公司开发了一个程序来检查KataGo的弱点。在玩了超过一百万场游戏后,它能够找到一个可以被一个体面的业余玩家利用的弱点。Pelrine用同样的方法击败了另一个顶级AI选手Leela Zero。
Here’s how it works: The goal is to create a large “loop(圈)” of stones(棋子) to surround an opponent’s group in a circle, and then distract the computer by making moves in other areas of the board. Even when his group was almost surrounded, the computer didn’t notice the strategy. “As a human, it would be pretty easy to spot,” Pelrine said, as the encircling checkers(环绕的棋子) stand out clearly on the game board. The bug shows that AI systems can’t really “think” beyond their training, so they often do things that look incredibly stupid to humans.
它是这样工作的: 目标是创造一个大的石头 “循环”,将对手的团队围成一圈,然后通过在电路板的其他区域移动来分散电脑的注意力。即使他的团队几乎被包围了,电脑也没有注意到这个策略。“作为一个人,很容易被发现,” Pelrine说,因为环绕的跳棋在游戏板上很明显。这个漏洞表明人工智能系统不能真正 “思考” 超出他们的训练,所以他们经常做一些对人类来说非常愚蠢的事情。
剩余段落暂不支持预览,下载文章即可阅读全文,有任何疑问请您联系客服【18011429143】微信同号。