利用AI抓贼，中国的AI技术已经领先世界

夏洛克，让道了，中国的神探们说可以利用人工智能（A.I.）抓住罪犯——你要不信，就请看他们如何在演唱会中抓住了一名偷土豆的贼。

今年5月，香港歌手张学友在华东的嘉兴举办演唱会，当地警方将A.I.应用于人脸识别，称从20000多观众中逮住一名土豆大盗。这位毫无疑心的疑犯刚通过演唱会的安保系统，不多久就被抓了：一项算法将他的脸与重点通缉犯头像数据库中的一张照片匹配起来。当局以盗窃价值17000美元土豆的罪名将其逮捕。

使用北京旷视科技研发的软件以来，这名盗贼已是张学友演唱会上被抓捕的第三名逃犯。而旷视科技是中国多家从事A.I.与人脸识别技术相结合产业的先驱企业之一。阿里巴巴旗下的移动支付平台蚂蚁金服利用一项“微笑支付”功能，使用户在肯德基点餐更便捷。杭州一所高中以此技术监控学生出勤率。深圳以及其他城市的交警则以之锁定乱穿马路的行人和自行车。还有北京天坛附近一处公园，将该技术应用于公厕，以防止游客盗窃厕纸。

这都反映了世界第二大经济体拥抱A.I.技术的非凡热忱。习近平主席承诺，到2030年，中国将成为人工智能领域的全球领导者，并以此创造价值近1500亿美元的国内产业。

面对中国的A.I.之梦，世界各国是否应该担忧？也许不必。不管是中国的分析人员还是美国的官员，其多数对于中国相关投入的评估都建立于一个共同的假设，就是该计划按照宣传内容切确落实。诚然，中国政府加大了对国有企业的扶持力度，对外企加以限制，并毫不吝啬对关键产业予以补贴，但中国的A.I.霸业还远没有一个定数。“A.I.不少难关本质上都是全球难题，”而“政府无法独自解决，”麦肯锡（McKinsey）于6月就相关专题发出的报告中写道。

谷歌大中华区（Google China）前总裁李开复强调，A.I.正由发现阶段向实施阶段转换，前者由美国主导，至于后者，中国享有重大“结构优势”。主要驱动力是什么？数据、计算能力，还有高水平的工程师。这些方面，人口第一大国占尽优势。

但人工智能的倡导者们发出告诫，数百万就业机会将会因此丧失，对于一个就业严重依赖重复性制造业的国家来说，前景不容乐观。中国将如何应对？看来，深度学习也会造成深度问题。（财富中文网）

本文另一版本刊载于2018年7月1日的《财富》杂志，标题为“微微破碎的‘黑镜’”。

译者：沈昕宇

STEP ASIDE, SHERLOCK. Detectives in China say they can catch criminals using artificial intelligence—and if you don’t believe them, consider the case of the potato thief at the pop concert.

Officials in the eastern Chinese city of Jiaxing in May used A.I.-powered facial-recognition technology to nab the alleged tater taker from a crowd of more than 20,000 people attending a performance by Hong Kong crooner Jacky Cheung. Moments after passing through the concert’s security system, the unsuspecting suspect was busted: An algorithm matched his face with an image from a database of “most wanted” mug shots. Authorities seized the man on charges of stealing $17,000 worth of potatoes.

The thief was the third fugitive to be arrested at a Jacky Cheung concert in as many months using software developed by Beijing’s Megvii, among the many Chinese groups pioneering ways to combine A.I. and facial-recognition capabilities. Alibaba Group mobile payments affiliate Ant Financial uses a “smile to pay” feature to facilitate purchases at KFC. A high school in Hangzhou monitors students’ attentiveness in class. Traffic police in Shenzhen and other cities spot jaywalkers and reckless bike couriers. A park near Beijing’s Temple of Heaven uses the technology in a public restroom to stop patrons from stealing toilet paper.

All of this hints at the extraordinary zeal with which the world’s second-largest economy has embraced A.I. President Xi Jinping vows China will become the global leader in artificial intelligence by 2030, creating a domestic industry worth nearly $150 billion.

Should the rest of the world be alarmed by China’s A.I. dreams? Perhaps not. Implicit in most assessments of the country’s efforts, whether by U.S. officials or Chinese analysts, is the shared assumption that the programs will perform as advertised. Though the Chinese government has certainly stepped up support for state-owned enterprises, tightened restrictions on foreign firms, and doled out massive subsidies to key sectors, his country’s future A.I. supremacy is far from guaranteed. “Many of the challenges of A.I. are global in nature,” reads a June report from McKinsey on the subject, and “not for government to solve alone.”

Kai-Fu Lee, the former head of Google China, argues that A.I. is shifting from a U.S.-led Age of Discovery to an Age of Implementation in which China enjoys significant “structural advantages.” The main drivers? Data, computing power, and competent engineers—all of which favor the world’s most populous nation.

Yet proponents of artificial intelligence warn that it could wipe out millions of jobs, a troubling prospect in a country that remains so heavily dependent on repetitive manufacturing jobs. How will China cope? Deep learning, it seems, can also raise deep questions.

A version of this article appears in the July 1, 2018 issue of Fortune with the headline “'Black Mirror,' Slightly Broken.”