人工智能发展如何,就藏在技术炒作曲线里
今年的 Gartner 技术炒作曲线并未得到太多媒体关注,所谓技术炒作曲线是 Gartner 对于技术发展周期的研究方法 ,这套方法论认为,每一项技术的生命周期都有五大阶段:
- Technology Trigger
- Peak of Inflated Expectations
- Trough of Disillusionment
- Slope of Enlightenme
- Plateau of Productivity
Gartner 通过五大阶段绘制的炒作曲线,主要针对投资人、创业者以及大公司的管理高层,是面向未来的技术发展蓝图,与普通用户的关系并不大。
下图是 2016 年的炒作曲线:
中文互联网世界直到 9 月底才有人跟进这个消息,这篇中文解读围绕各个技术要点进行了非常深入的分析,非常值得认真阅读,尤其是在当下人工智能概念甚嚣尘上的时候,你应该了解通用机器智能(General-Purpose Machine Intelligence)的发展情况:
机器具有类似人类在学习、推理、适应和理解等方面的能力就被称为“智能”。通用机器智能(GPMI协议;又名“强人工智能”)适用于广泛的使用案例,专用机器智能(“弱人工智能”)仅限于特定(窄)的使用案例。GPMI协议对智能机器不是必要的,具有能够控制自身维护和再生产方式的具有GPMI的机器只存在于科幻小说。今天人工智能系统,都不能通过等同于人类的智力的通用测试(也没有完全可接受对人类的智慧的测量方式去测试机器)。这并不是说,永远不可能创造一种接近于人类认知能力的机器,但我们可能距离完成必要研究和工程实验还有好几十年。
MailChimp 是一家怎样的奇葩公司
在很多人的印象中,一个硅谷科技创业公司的流程是这样的:
出身名校的创始人用一个 idea 得到投资人的青睐,然后通过一个视频在众筹网站上获取第一批用户支持,接着通过某个创业加速器,比如 Y Combinator,获取更多外部支持,从而快速展开业务。
但 MailChimp 却是一家与硅谷气质完全不同的科技公司,它没有拿过风险投资的钱,《纽约时报》专栏作者 Farhad Manjoo 写道:
Under the radar, slowly and steadily, and without ever taking a dime in outside funding or spending more than it earned, MailChimp has been building a behemoth. According to Ben Chestnut, MailChimp’s co-founder and chief executive, the company recorded $280 million in revenue in 2015 and is on track to top $400 million in 2016. MailChimp has always been profitable, Mr. Chestnut said, though he declined to divulge exact margins. The company — which has repeatedly turned down overtures from venture capitalists and is wholly owned by Mr. Chestnut and his co-founder, Dan Kurzius — now employs about 550 people, and by next year it will be close to 700.
在其创始人看来,创业公司融资后的一个问题就是容易养成乱花钱的坏习惯。事实上,这个世界上并非所有的生意都需要投资,由于资本的逐利性和短视,创业公司、创业者很容易被资本绑架,尤其是在中国当下急躁的创业环境中,令我印象深刻的一个案例,去年一位 2B 的软件公司创始人对竞争对手上市新三板非常不屑,他当时告诉我:「我从来不觉得新三板是上市。」但不到一年,这家公司也成为新三板的上市公司……
书摘:亚马逊 「One Click」专利背后
坦率来说,Richard Brandt 写的《一键下单:杰夫·贝佐斯与亚马逊的崛起》并不好,这本书的英文版出版于 2011 年,彼时,亚马逊帝国还未形成,其商业技术影响力远不如当时如日中天的苹果,也很少有人真正重视亚马逊持续不断烧钱搞得云计算。但 Richard Brandt 书中对于 「One Click」,也就是亚马逊的「一键下单」的叙述很精彩,下面是这本书英文版的摘录:
“The 1-Click software was written mostly by a programmer named Peri Hartman, who joined Amazon in 1997. Hartman was given the task of working on the software that would be the interface to the customers, including the ordering system they would use to buy books.
Hartman recalls that, one day over lunch with Bezos and Shel Kaphan, the head of software development, Bezos told them, “We need something to make the ordering system frictionless. We need to make it so the customer can order products with the least amount of effort. They should be able to click on one thing, and it’s done.”
A simple idea on simplicity, and Hartman took it literally. He set to work creating a program that would require just one click to order a product. (His name appeared first on the patent application, which became patent number US5960411.) The patent filing was titled, innocuously enough, “Method and system for placing a purchase order via a communications network.” More telling was the label in an illustrated flow chart in the patent: “Enable Single-Action Ordering.”
“In fact, it is a simple idea. The nineteen-page patent filing is made up mostly of flow chart diagrams that show the sequence of events that enable buyers to place an order with a single click: Retrieve the buyers’ identification and payment methods when they first place an order, enter it into the system, and the next time they look up a book, they’ll see a 1-Click ordering button on the buyers’ pages.
Sound like a no-brainer? That’s what Amazon’s competitors thought. Many people in technology hate this type of broad patent, known as a “process patent” since it mainly describes the process of doing something that is, well, patently obvious—in this case, reducing to one the number of clicks needed to make a purchase.
“That is, however, the kind of attention to detail that has helped make Amazon.com a success. Jeff Bezos will do anything he can think of to make the process of using Amazon.com easier. The genius is that Bezos thought of it first. He knew that doing anything online had to be simple in an age when people were becoming overwhelmed by complicated computers, software, and Internet technology. This, in fact, was a principle that Larry Page and Sergey Brin adopted a couple years later when creating Google. But few competitors have had the sense to follow this rule. Even today, most Web sites seem to be confusing morasses of text, graphics, videos, flashing ads, and a tangled string of links. Amazon.com doesn’t match the simplicity of Google’s famously spare site, but it is well designed and easy to navigate.”