IBM超级计算机“沃森”转战商场

    IBM公司鼎鼎大名的超级计算机沃森（Watson）在《危险边缘》（Jeopardy，美国一档著名的智力竞赛节目——译注）节目中击败了低劣的人类，但这仅仅是它小试牛刀的起点而已。很快，它将作为商用分析工具，在医疗、电信和金融服务等数据密集型行业大显身手。

    一年前，沃森就已开始追随前辈、棋坛高手“深蓝”（Deep Blue）的脚步，证明机器比人类聪明（假如有人怀疑这一点的话）。现在，沃森迎来了它的首位客户——位于印第安纳波利斯市的伟彭公司（WellPoint）。这家保险公司准备将这台能量非凡的分析机器投入试运行。据IBM公司“沃森解决方案”（Watson Solutions）部门的总经理马诺伊•萨杰那介绍，未来数月，更多客户将会跟进启用沃森系统，它们涵盖了从医疗到金融服务业的广泛领域。

    多年来，IBM一直致力于打造一台能处理自然语言的计算机，它能又快又准地回答各种问题。沃森正是多年努力的结晶。这个计算集群由90台服务器组成，去年在《危险边缘》节目中首度亮相。但是，它可绝不是一台仅供游戏竞赛的小玩意儿。IBM公司计划将沃森作为一项依托于云计算的服务进行销售。借助这一产品，用户可在不同的数据集里找到各种问题的答案。比如，金融服务公司能用它在新闻报道和市场研究报告中进行筛选，锁定潜在的收购目标。医疗保健公司则能运用沃森处理医疗论文、历史病历甚至是病人自己的病史记录，由此找出最对症的诊断，确定最佳治疗方案。

    当然，数据分析并不是什么新鲜事，但当今世界上所产生的海量数据却是前所未有的事物。它们排山倒海，毫无条理，涵盖了从Twitter留言、Facebook更新，到亚马逊网站（Amazon）上的评论等在内，五花八门的信息。据萨杰那称，现在全球90%的信息是在过去两年间产生的。炮制数据不费吹灰之力，理解数据去也困难得多。这正是沃森和其他新一代分析型计算机应运而生的原因。超级计算机能在三秒内处理2亿个网页。此外，它还能理解人类的语言。可以将它想象为超大版本的iPhone语音助手Siri，即苹果公司（Apple）的智能个人助理软件。实际上，萨杰那表示，将来某一天，用户确实可以智能手机或平板电脑上直接使用沃森的服务。他还计划“教”沃森各种语言，包括日语和法语，以便为海外客户服务。

    Beating lowly humans on Jeopardy was just the beginning. IBM's famed Watson supercomputer will soon be available as a commercialized analytics tool for data-heavy industries like healthcare, telecom and financial services.

    It's been one year since Watson followed in the footsteps of its chess-playing predecessor Deep Blue and proved -- just in case anyone had a doubt -- that machines are smarter than people. Now, Watson's first pilot customer, Indianapolis-based insurance company WellPoint (WLP), is getting ready to take the massive analytics engine for a test drive. More beta customers (spanning both healthcare and financial services) will follow in the next few months, according to IBM's (IBM) general manager of Watson Solutions, Manoj Saxena.

    Watson was the result of IBM's years-long quest to build a natural language processing machine that could answer questions with speed and accuracy. The computing cluster, comprised of 90 servers, made its debut on Jeopardy last year. But it wasn't just a game show gimmick. IBM plans to sell Watson as a cloud-based service companies can tap to find answers in disparate data sets. For example, a financial services firm could use it to sift through news reports and market research to find likely acquisition targets. Or a healthcare company could utilize Watson to process medical articles, prior cases and even a patient's own medical history and identify the most likely diagnosis and best course of treatment.

    Of course, data analytics is nothing new. But the terabytes upon terabytes of unstructured data in the world (including Tweets, Facebook updates and Amazon (AMZN) reviews) is unprecedented. According to IBM's Saxena, 90% of the world's information was generated in the last two years. Producing data is much easier than making sense of it. That's where Watson and other next-generation analytics tools come in. The supercomputer can process 200 million web pages in three seconds. What's more, it can understand human language. Think of it as the iPhone's Siri, Apple's (AAPL) intelligent personal assistant, on intergalactic steroids. In fact, Saxena says users will someday be able to access Watson from their smartphone or tablet. He also plans to "teach" Watson different languages, including Japanese and French, so it could work for customers abroad.