|
How Europe can improve the development of AI Its real clout comes from its power to set standards Print edition | Leaders Sep 20th 2018 THE two superpowers of artificial intelligence (AI) are America and China. Their tech giants have collected the most data, attracted the best talent and boast the biggest computing clouds—the main ingredients needed to develop AI services from facial recognition to self-driving cars. Their dominance deeply worries the European Union, the world’s second-largest economic power. It is busily concocting plans to close the gap. 美国和中国是人工智能的两个超级大国，它们的科技巨头收集了最多的数据，吸引了最多的人才，并拥有最大的计算云——从面部识别到自动驾驶汽车，云计算是开发人工智能服务所需的主要要素。它们的主导地位令世界第二大经济体欧盟深感担忧，因此欧盟也在盘算着如何缩小与他们的差距。 That Europe wants to foster its own AI industry is understandable. Artificial intelligence is much more than another Silicon Valley buzzword—more, even, than seminal products like the smartphone. It is better seen as a resource, a bit like electricity, that will touch every part of the economy and society. Plenty of people fret that, without its own cutting-edge research and AI champions, big digital platforms based abroad will siphon off profits and jobs and leave the EU a lot poorer. The technology also looms large in military planning. China’s big bet on AI is partly a bet on autonomous weapons; America is likely to follow the same path. Given the doubt over whether America will always be willing to come to Europe’s defence, some see spending on AI as a matter of national security. 欧洲希望开发自己的人工智能产业是可以理解的。人工智能远远超过硅谷的另一个流行词——甚至超过智能手机等开创性产品。它更像是一种资源，有点像电力，它将涉及经济和社会的方方面面。许多人担心，如果没有自己的前沿研究和人工智能重磅成果，总部设在海外的大型数字平台将会吸走欧盟的利润和就业机会，让欧盟变得更加贫穷。这项技术在军事规划中也占有重要地位。中国对人工智能下了大赌注，在一定程度上是押注自动武器；美国很可能会走上同样的道路。考虑到也许美国不会始终愿意参与欧洲防务，欧盟的一些人认为在人工智能上的支出事关国家安全。 Both arguments make sense. But can Europe support AI without wasting money or lapsing into protectionism? The EU has a dismal record in high-tech industrial policy. Witness Quaero, a failed attempt to build a European alternative to Google, or the Human Brain Project, which has spent over €1bn ($1.17bn) with little to show for it. Experts warn against the rise of “AI nationalism”, whereby countries increasingly try to keep their data and their algorithms to themselves. 这两种观点都有道理。但欧洲发展人工智能可以避免不浪费资金或陷入保护主义吗?欧盟在高科技产业政策方面并不乐观，比如Quaero，欧洲取代谷歌或建立人类大脑计划的失败尝试，已经花费超过10亿欧元（11.7亿美元），但收获甚微。专家们对“人工智能民族主义”的兴起发出了警告，基于此，越来越多的国家将数据和算法保密。 Two aims should guide EU policy. Instead of focusing its financing on high-profile individual projects, Europe should create the environment for its AI industry to thrive. And instead of keeping foreign providers out, it should use its clout to improve their behaviour. 有两个目标可以指导欧盟的政策。欧洲应该为其人工智能行业的蓬勃发展创造环境，而不是将资金集中在受关注的个别项目上。与其把外国供应商拒之门外，还不如利用自己的影响力来改善他们的行为。 Creating the right environment means, above all, working to overcome the fragmentation that bedevils Europe. Big and homogeneous home markets give America and China the huge advantage of scale. According to one estimate, China will hold 30% of the world’s data by 2030; America is likely to have just as much. Europe has data, too, but needs to pool its resources. To its credit, the European Commission is arguing for a common market for data. But much more needs to be done, such as laying down rules about how data held by companies and governments can be shared. 创造合适的环境首先意味着要努力克服一直困扰欧洲的分裂。庞大而均质的国内市场给了美国和中国巨大的规模优势。据估计，到2030年，中国将占据世界数据的30%，美国很可能拥有同样多的资源。欧洲也有数据，但需要集中资源。值得称赞的是，欧盟委员会主张建立一个共同的数据市场，但还有很多工作要做，比如制定关于公司和政府持有的数据如何共享的规则。 National faultlines also cut deep in research and development. Germany has downgraded plans to co-operate with France in AI research, for example. In addition, Europe’s existing research bureaucracy is adept at sucking up funds, to the detriment of startups and outsiders. Better to encourage grass-roots initiatives such as CLAIRE and ELLIS, which seek to create Europe-wide networks of research labs. France has launched JEDI, short for Joint European Disruptive Initiative, an attempt to mimic America’s Defence Advanced Research Projects Agency (DARPA), which allocates money using open competitions and does not hesitate to cull programmes that fail to show promise. More opportunities of this sort, plus an accommodating immigration regime, would attract and retain AI researchers, who often decamp to America (and sometimes even to China). 国家在研究和开发上也有很深的断裂层。例如，德国已经降低了与法国在人工智能研究领域合作的计划。此外，欧洲现有的研究机构善于吸收资金，损害初创企业和外来企业的利益。更好的做法是鼓励像克莱尔和埃利斯这样的草根倡议，他们寻求建立全欧洲范围的研究实验室网络。法国启动了“欧洲联合颠覆计划”(简称JEDI)，旨在模仿美国国防高级研究计划局(DARPA)的做法，通过公开竞争来分配资金，果断剔除那些没有前途的项目。更多这样的机会，再加上通融的移民制度，将吸引并留住人工智能研究人才，这些人通常会逃往美国(有时甚至是中国)。 European policymakers can also make better use of the one area where they are world-class—setting standards. Europe’s market of 500m relatively wealthy consumers is still enticing enough that firms will generally comply with EU rules rather than pull out. An example is a strict new privacy law, the General Data Protection Regulation; the principles of the GDPR are now being used as a benchmark for good data practice in markets well beyond Europe. By imposing common rules, such standards can help the EU’s indigenous AI industry flourish. But they could also have a more subtle effect—of making AI from outside the EU more benign. 欧洲政策制定者还可以更好地利用这块区域，在这块区域，他们制定世界一流标准。欧洲市场有5亿相对富裕的消费者，这一市场仍然具有足够的吸引力，因此企业一般会遵守欧盟的规定，而不是撤出。一个例子是严格的新隐私法——通用数据保护条例（GDPR）;GDPR原则被用作良好的数据实践基准，其应用范围远超欧洲。通过实施共同的规则，这些标准可以帮助欧盟本土人工智能产业蓬勃发展。但它们也可能产生更微妙的效果——让来自欧盟以外的人工智能变得更加友好。 By the rule book America and China both represent flawed models of data collection and governance. China sees AI as a powerful tool to monitor, manage and control its citizens. America’s tech titans scoop up users’ data with insufficient regard for their privacy. The GDPR is just the start. Robust standards are needed to ensure that AI services are transparent and fair and that they do not discriminate against particular groups. Europe has a chance to shape the development of AI so that this vital technology takes more goals into account than simply maximising advertising income and minimising dissent. Even if it comes up with policies that help its native AI industry thrive, Europe may never match America and China. But it can nonetheless help guide AI onto a path that benefits its own citizens, and those in the rest of the world. 美国和中国都代表着有缺陷的数据收集和管理模式。中国将人工智能视为监控、管理和控制本国公民的强大工具。美国的科技巨头们在没有充分考虑用户隐私的情况下获取了用户的数据。GDPR只是一个开始。为了确保人工智能服务的透明度和公平性，以及不歧视特定群体，需要制定健全的标准。欧洲可以打造人工智能的发展，使这项至关重要的技术能够考虑更多的目标，而不仅仅是最大化广告收入和最小化异议。即使欧盟提出了帮助本土人工智能产业繁荣发展的政策，欧洲也可能永远无法与美国和中国匹敌。尽管如此，它仍能帮助引导人工智能走上一条有益于自己和世界其他地区公民的道路。 This article appeared in the Leaders section of the print edition under the headline "AI, EU, go" 英文部分选自《经济学人》杂志
|
|
|
