双语:Artificial Intelligence: Million-dollar Babies
发布时间:2017年10月04日
Economist 译  

Artificial Intelligence: Million-dollar Babies

人工智能:百万美元宝贝

 

As Silicon Valley fights for talent, universities struggle to hold on to their stars

硅谷抢夺人才,大学难留明星学者

 

That a computer program can repeatedly beat the world champion at Go, a complex board game, is a coup for the fast-moving field of artificial intelligence (AI). Another high-stakes game, however, is taking place behind the scenes, as firms compete to hire the smartest AI experts. Technology giants, including Google, Facebook, Microsoft and Baidu, are racing to expand their AI activities. Last year, they spent some $8.5 billion on research, deals and hiring, says Quid, a data firm. That was four times more than in 2010.

 

计算机程序可以反复战胜围棋世界冠军,这是人工智能这一快速发展的领域中一项极为难得的成就。然而,随着各家公司竞相把顶尖的人工智能专家招致麾下,另一场高风险游戏正在幕后展开。包括谷歌、Facebook、微软、百度在内的科技巨头争相扩展其人工智能项目。数据公司Quid表示,去年,这些科技公司花费了约85亿美元用于研究、收购及网罗人才,比2010年多四倍。

 

In the past universities employed the world’s best AI experts. Now tech firms are plundering departments of robotics and machine learning (where computers learn from data themselves) for the highest-flying faculty and students, luring them with big salaries similar to those fetched by professional athletes.

 

过去,大学拥有世界一流的人工智能专家。如今,科技企业正从大学的“机器人及机器学习(计算机通过数据自动学习)”系里抢夺优秀师生,以堪比职业运动员的高薪做诱饵。

 

Last year Uber, a taxi-hailing firm, recruited 40 of the 140 staff of the National Robotics Engineering Centre at Carnegie Mellon University, and set up a unit to work on self-driving cars. That drew headlines because Uber had earlier promised to fund research at the centre before deciding instead to peel off its staff. Other firms seek talent more quietly but just as doggedly. The migration to the private sector startles many academics. “I cannot even hold onto my grad students,” says Pedro Domingos, a professor at the University of Washington who specialises in machine learning and has himself had job offers from tech firms. “Companies are trying to hire them away before they graduate.”

 

美国卡耐基梅隆大学的国家机器人工程中心原本有140名教师,去年,打车公司优步从中招聘了40人,设立部门研究自动驾驶汽车。此举惹来关注,因为优步之前承诺资助该中心的研究工作,后来却转而挖角。其他公司寻觅人才的举动则相对低调,但也同样执着。人才向私营公司的流动让不少学者感到震惊。“我连自己的研究生也留不住,”华盛顿大学的佩德罗·多明戈斯教授说道,他是机器学习方面的专家,连他自己也收到了科技公司伸出的橄榄枝,“学生还没毕业,那些公司就想把他们聘走。”

 

Experts in machine learning are most in demand. Big tech firms use it in many activities, from basic tasks such as spam-filtering and better targeting of online advertisements, to futuristic endeavours such as self-driving cars or scanning images to identify disease. As tech giants work on features such as virtual personal-assistant technology, to help users organise their lives, or tools to make it easier to search through photographs, they rely on advances in machine learning.

 

机器学习领域的专家最为抢手。大型科技公司的许多任务都要运用这一技术,从一些基本任务,如过滤垃圾邮件和令网络广告更有针对性,到无人驾驶汽车或扫描图像来发现疾病等具有未来色彩的尝试,无一例外。科技巨头在研发一些产品时要依赖机器学习技术的进步,比如帮助用户安排生活的虚拟个人助理或是方便人们搜寻图片的工具。

 

Tech firms’ investment in this area helps to explain how a once-arcane academic gathering, the Conference on Neural Information Processing Systems, held each December in Canada, has become the Davos of AI. Participants go to learn, be seen and get courted by bosses looking for talent. Attendance has tripled since 2010, reaching 3,800 last year.

 

科技公司对这一领域的投资有助解释为何“神经信息处理系统大会”(每年12月在加拿大举行)这一曾被视为高深莫测的学术会议如今摇身成为人工智能界的达沃斯盛会。与会者一方面为了学习,另一方面也为了被求贤若渴的老板们发现并追捧。2010年以来,其与会人数增加了两倍,去年达到3800人。

 

No reliable statistics exist to show how many academics are joining tech companies. But indications exist. In the field of “deep learning”, where computers draw insights from large data sets using methods similar to a human brain’s neural networks, the share of papers written by authors with some corporate affiliation is up sharply.

 

学术界有多少人转投科技公司的怀抱目前仍无可靠统计数据,但有迹可循。“深度学习”是指计算机利用近似人类大脑神经网络的运作方式从大型数据集中析取知识,这一范畴的学术论文中,在企业任职的作者比例大幅上升。

 

Tech firms have not always lavished such attention and resources on AI experts. The field was largely ignored and underfunded during the “AI winter” of the 1980s and 1990s, when fashionable approaches to AI failed to match their early promise. The present machine-learning boom began in earnest when Google started doing deals focused on AI. In 2014, for example, it bought DeepMind, the startup behind the computer’s victory in Go, from researchers in London. The price was rumoured to be around $600m. Around then Facebook, which also reportedly hoped to buy DeepMind, started a lab focused on artificial intelligence and hired an academic from New York University, Yann LeCun, to run it.

 

科技公司并非一开始就对人工智能专家倾注如此多的心思和资源。在上世纪八九十年代的“人工智能寒冬”,新潮的人工智能技术未如预期,该领域被广为忽视,资金投入也不足。目前这股“机器学习”热潮是在谷歌开始收购专注人工智能技术的公司后才真正开启的。比如,2014年,谷歌从伦敦的研究人员手中收购了DeepMind,这家创业公司正是人机围棋大战中计算机取胜的幕后关键。据传当时的收购价约为六亿美元。据报道也曾有意收购DeepmMindFacebook也在差不多同一时间建起实验室,专注研发人工智能技术,并从纽约大学请来学者燕乐存来做负责人。

 

The firms offer academics the chance to see their ideas reach markets quickly, which many like. Private-sector jobs can also free academics from the uncertainty of securing research grants. Andrew Ng, who leads AI research for the Chinese internet giant Baidu and used to teach full-time at Stanford, says tech firms offer two especially appealing things: lots of computing power and large data sets. Both are essential for modern machine learning.

 

这些公司为学者们提供机会,让其创意迅速推向市场,往往大受欢迎。私营公司的职位也令学者们不用担心研究经费不足的问题。之前在斯坦福大学全职任教的吴恩达目前效力于中国互联网巨头百度,主管人工智能研究。他表示,科技公司能提供两个特别诱人的条件:强大的计算能力和庞大的数据集。这两者为现代机器学习研究必不可少。

 

All that is to the good, but the hiring spree could also impose costs. One is that universities, unable to offer competitive salaries, will be damaged if too many bright minds are either lured away permanently or distracted from the lecture hall by commitments to tech firms. Whole countries could suffer, too. Most big tech firms have their headquarters in America; places like Canada, whose universities have been at the forefront of AI development, could see little benefit if their brightest staff disappear to firms over the border, says Ajay Agrawal, a professor at the University of Toronto.

 

这些都是好的方面,但挖角热潮也有代价。一方面,大学由于无法提供具有竞争力的薪酬,假如过多优秀人才被诱走,一去不返,或是忙于服务科技公司而无法专心讲学,大学将蒙受损失。同时,一些国家也可能遭罪。大型科技公司总部多在美国;像加拿大这样的国家,其大学一直处于人工智能研发的前沿,如果他们最聪明的人才都被境外公司吸引走,对本国实在毫无益处,多伦多大学的阿杰伊·阿格拉沃尔教授说道。

 

Another risk is if expertise in AI is concentrated disproportionately in a few firms. Tech companies make public some of their research through open sourcing. They also promise employees that they can write papers. In practice, however, many profitable findings are not shared. Some worry that Google, the leading firm in the field, could establish something close to an intellectual monopoly. Anthony Goldbloom of Kaggle, which runs data-science competitions that have resulted in promising academics being hired by firms, compares Google’s pre-eminence in AI to the concentration of talented scientists who laboured on the Manhattan Project, which produced America’s atom bomb.

 

另一风险是人工智能技术过度集中于少数企业手中。科技公司通过开源方式公开其部分研究成果。它们也答应员工可以撰写论文。然而,实际上,许多有利可图的研究成果并未共享。有人担心,作为人工智能界领头羊的谷歌可能形成近乎知识垄断的地位。Kaggle是组织数据学竞赛的平台,不少公司通过这些比赛搜罗学术新星,该平台的安东尼·古德鲁姆将谷歌在人工智能上的卓越表现与当年集结众多科学英才在曼哈顿计划中努力工作相提并论。该计划最终为美国造出原子弹。

 

Ready for the harvest?

准备收获?

 

The threat of any single firm having too much influence over the future of AI prompted several technology bosses, including Elon Musk of Tesla, to pledge in December to spend over $1 billion on a not-for-profit initiative, OpenAI, which will make its research public. It is supposed to combine the research focus of a university with a company’s real-world aspirations. It hopes to attract researchers to produce original findings and papers.

 

由于担心某一家企业在未来对人工智能的发展拥有过大影响力,多家科技公司的老板,包括特斯拉的伊隆·马斯克在去年12月承诺向非营利研究机构OpenAI提供10亿美元资助。这一机构将公开其研究成果。该机构应该能结合大学的研究热点和企业的现实抱负,希望能吸引研究人员做出原创成果及论文。

 

Whether tech firms, rather than universities, are best placed to deliver general progress in AI is up for debate. Andrew Moore, the dean of Carnegie Mellon University’s computer-science department, worries about the potential for a “seed corn” problem: that universities could one day lack sufficient staff to produce future crops of researchers. As bad, with fewer people doing pure academic research, sharing ideas openly or working on projects with decades-long time horizons, future breakthroughs could also be stunted.

 

最有优势推动人工智能研究总体进展的是科技企业而非高等院校么?众说纷纭。卡耐基梅隆大学计算机系系主任安德鲁·摩尔担心“希望种子”成忧:大学终有一天会缺乏教师培养未来的研究人员。同样糟糕的是,越来越少人会从事纯学术研究、公开分享想法或者参与跨度达几十年的研究项目,未来的突破也可能受到制约。

 

But such risks will not necessarily materialise. The extra money on offer in AI has excited new students to enter the field. And tech firms could help to do even more to develop and replace talent, for example by endowing more professorships and offering more grants to researchers. Tech firms have the cash to do so, and the motivation. In Silicon Valley it is talent, not money, that is the scarcest resource.

 

但忧虑未必会成真。投入人工智能研发的额外资金激励新生步入这一领域。而且科技公司可以进一步推动人才发展及更新换代,比如授予更多教授职位,向研究人员提供更多资助。科技公司有财力也有动力这么做。在硅谷,最稀缺的资源是人才,而非金钱。


下载:中文、英文版本