Kai-Fu Lee: I see the future of science and technology in Silicon Valley-Google, Android,VR-IT information

15 day, 100 people, beginning of the year 2016, Lee himself headed to Silicon Valley.

26 Ding Ding heads, DST Milner, Google Pi Cai, Yahoo Jerry Yang, and YC incubator SAM, father of Android Andy Rubin ... ...

18 among the most prestigious companies, Apple, Google, Facebook, Airbnb, Tesla, Twitter, YC incubators ... ...

Science and technology innovation in San Diego what trends are happening in Silicon Valley? Next phase sweeping the globe, what kind of technology will change the world? Through investment lays the current DST founding partner James Milner in the pattern of the Internet how to find the "billion-dollar company" founder? Father of Android Andy Rubin was to repeat his Android miracle? Upgrade Google what ambitions behind the Alphabet?

Recently, China's "Silicon Valley", Zhongguancun, Kai-Fu Lee to the media, shares his new year "stories of Silicon Valley." Stories include his current Chinese entrepreneurs thoughts and recommendations, Silicon Valley's most cutting-edge technology, ongoing technology trends, combined with Google's strategy for the foreseeable future, as well as think about VR and AR hot this year.

Following is Kai-Fu Lee dictating consolidation:

1, robot

To talk about some of the more eye-catching things. I think it has some interesting discussions at each meeting. For example with Andy Rubin (father of Android) very interesting discussions, his company called Playground Global, in fact, much has been reported on the Wall Street Journal. See him that day, we saw the Boston Dynamics (Google's large military robotics firm Boston Dynamics), the robot dog.

This dog is very easy to operate the machine, I also play a bit. Say it was fate, because Boston Dynamics Founder Marc Raibert, up from CMU (Media Note: Carnegie Mellon, Kai-Fu Lee alma mater) Professor when I was a student, his Office was next door to me. When his project at CMU is a hopping and jumping robots fall down for a minute. However, at that time, if you gently touched with a stick, it will fall down. And there were a bunch of very thick and very long lines, connections on the computer, this is the first case.

Few days ago, the kick will not be crowding the screen robot dogs and robots is already iteration has improved a lot. Marc Raibert was founded Boston Dynamics, and won the United States Department of defense research funding, specializes in robotics research . From the foot to four feet, then made it back to two feet, each step is not easy, so it took United States Department of Defense to close billions of dollars of money. Then Google after it bought to Boston Dynamics, buy Google didn't make it again after the Ministry of defence.

Buy Marc Raibert Boston Dynamics of decision makers at the time was Andy Rubin. Because after made Android Andy Rubin, Google hopes Chrome and Android and up, by Sundar Pichai (Google CEO Pi Cai) take over his management. And then created the GoogleX, please go to the Google x Andy Rubin do robots, because his own hobby is robots. And the Android is mobile robot. That day we communicate with Andy Rubin, Marc Raibert visit Andy Rubin with a robot dog, so we see it. It's all about Google robot dog and robot I know of.

2, robot platform

Andy Rubin later left the Google x, founded a company called Playground Global. In short, his company was hoping to make a robot version of the Android platform. If mobile phone is any guide, about ten years ago, if you want to open a cell phone company may have to spend $ 100 million or 200 million doing research and development, can be made to phone, but now you could spend $ 1 million to make. Because you can have a variety of OEM, there are standard modules and software with Android, there are others. What special features if you don't, you're going to have a cell phone. For now, create phone the hardware cost has been lowered 100 times, and popular. So you look now considered to represent the Internet cell phone manufacturers are all went out. Of course, the millet is when you start doing cell phone is quite expensive, but getting cheaper after.

Similarly, Andy Rubin said, are bound to occur if robots are to popularize this day, so he has done is to reduce the robot business modules such as threshold. Android provides a module of the smart phone, a mobile from mobile hardware until software becomes easier, and cost reduction, so that more people came in, this can allow more people to venture around the mobile phone, or venture too high a threshold .

Andy Rubin now think that robots are also smartphones under nearly ten years ago, so I made a robot, so that more people to do the robot business, it's going to be a thing to change the world, is also one of his childhood dreams. So the Playground Global is an Android for robot, but it's not just kind of walking robots. If you want to make a simple industrial robot, or a robot like that sweep the IFA, are available. From the perspective of robot modules, the robot is basically a bunch of sensors, combined training and learning and control, so that it can move – hands, feet, moving its claws or something like that. Then joined in the other parts of the system, such as input and recognition.

That's Andy Rubin shared his dream with us. His model and innovation works very similar to the early days, basically, is an incubator that is deeply involved in the project, and then to the valuable mining module for standardization. Then split out good project, as a separate company.

Of course he and our discussions in China have the opportunity to do a Playground, he had a number of investments in China, media reported. As to whether the future will cooperate and innovation works, we don't talk. In short, we admire his vision, but he has to do something, because the hardware, the complexity will increase many, this may be a challenge.

But as I see it, his venture, with millet is somewhat similar to a certain extent. Not saying he did a similar, but doing this requires a great deal of energy, needs to have a founder can indefinitely inhale, and investment of funds, and then to build things that others think it is difficult. In this regard, I feel that Lei has a similar place with him. Because he also has a large number of famous investors and involved is huge and giant high valuations, making a special long-term, storm is very difficult and risky design.

Another very interesting thing, and also a bit of sarcasm in it. Andy Rubin's Office now at Fry's Electronics, this is where geeks in Silicon Valley for years to buy parts, and Zhongguancun, but Fry's Electronics is a unified marketplace, only. Earlier, when geeks like us at the weekend would go there to see, is there anything cheaper PC Board, and then you can buy a home together, or hard disk is not at home, go over there to see if a discount, wholesale markets in short is a variety of electronic components.

But later, as e-commerce, we buy things won't go to the store, FRY's electronics sold half of the floor to Andy Rubin. He said his "ambition", he said Fry's Electronics must be closed down, I took his half of the floor, and it closed down, pour a layer I get a layer.

So Andy Rubin is very interesting, there are some bonus: at the forefront of the field, in the areas of Robotics and intelligent hardware, how are we going to repeat the miracle of Android.

3, artificial intelligence

Another interesting phenomenon is to do deep learning artificial intelligence doctoral student, now a graduate can get 200 to offer $ 3 million in income, it's never happened. I certainly agree that Silicon Valley companies are in hot pursuit of this direction , and the base is four major universities: Stanford, MIT, CMU, Berkeley. Doctoral students in these schools in Silicon Valley before you can get paid, but never to the salary level of $ 2.5 million per year.

Why would this happen? First, because people who really understand the depth of learning is not a lot, so the imbalance between supply and demand .

Second, is because it's value . Google got such a person, he will be able to immediately use the money he earns 100 times. Because you have such a person in a certain field, for example, if Google want to use cash on hand to make the best of his financial investments in the secondary market, a person earning 100 times a year, so this is definitely a bargain.

Third, because of competition . Google does not want such a person falling into the arms of his rivals. Because Google has a leading edge, but if this person went to Facebook, Microsoft, will give both the opportunity right away. So now is a war on talent. For this group of PhD graduates may be less than 50 a year, the big three: Google, Facebook and Microsoft, are used to dig at unreasonable prices.

This gave me two revelations. Pity on the one hand he was born too early, I was in this area, but there was no company to dig us. The other hand is similar to the talent wars, it might also be happening in China.

Dr I'm afraid no such 50 Chinese universities, but we have a few "University" graduate, is also the same as Silicon Valley. This "University" named Tencent, Baidu and Alibaba. So these people will in the future because of the lure of data, or loyalty to the company left to do something about it? Or is it dug between several companies, could become a very interesting thing.

Because I was doing it, so I think the depth of learning very much, but not so great. You make a smart man for two years, he can have this value. So in other words, we have experts to help train the depth of 1000 Chinese learning, these are very interesting discussions.

And then I asked them, you compete so hard, another two or three years, Chinese studies these things too hard, Chinese data than you, you have this open, Chinese talent in this field may be smaller than the United States.

In China, Baidu, Tencent, SINA and more enormous amounts of data, if China has $ 2.5 million annual salary of temptation, people frantically rushed to this area. So I believe that the final should be a State of Sino-American leader in the field, although China is far behind the United States, after all, the United States has Stanford, Berkeley, CMU, MIT out of people, and is constantly flowing. But for China, these people may be Baidu, Tencent and Ali. Baidu, Tencent and Ali himself now hold for these people, but they can never hold for long. But long term, millet, qihoo 360 also has such a person, so it's going to be a very interesting competition status in the industry.

In this, leading British Hoffman referred to musk and Peter who tyre set up an open source platform Open AI. This Open AI was founded mainly to prevent Google, Facebook and Microsoft companies a monopoly, you have so many computers, so rich, and dig into the best of them all, so we're going to make an open-source platform, to make sure that this thing can be more people learn quickly. So this is a very interesting thing, was very advanced in Silicon Valley.

4, VR and AR

Overall, on all AR VR, mainly feel there are so few.

For the VR, for development in this area, a group of people are very optimistic, while another group of people is a cycle--think we said that is still groping around, because good content is enough, experienced enough, is too expensive, then you want to connect to a PC, pull a thread, might really be in an age of players.

Innovation works, we integrated a number of observations. Our view is that the VR longer term impact on society should be particularly large, but Moore's role, it should be growing in bright, more do not dizziness, no line was doing more, grow light, grow smaller. Eventually becoming a, maybe not the glasses, but a certain pattern, so that before you know it, I put it into your life, it will come a day when, is a huge industry, will change all things.

But we may be a little more conservative. We may on the future of the 5-10, has a very optimistic expectation. But for what he can within one or two years, Oculus, HTC's products, they can break a player within a year or two areas can achieve universal status, we may still adopt a watch-and-grain of salt .

I certainly agree that from an investment perspective, now see the good companies have voted, voted only because you cannot enter the cycle, except to say that if we are to high expectations, says it will subvert anything, we still have a little time.

There is from the application level, VR SGI done before my own, do the 3D stuff. So I personally think VR is, it must was first breakthroughs in entertainment. Because we talked so much 3D stuff, are not verified, and only the demand for 3D entertainment content-related areas has been verified, so basically to the movies more, play more cool games, then became more lifelike, was such a State. Kind of like an extension of the film and gaming industries, but it is important to note, this is a great extension.

AR, it is possible to have different fields of application. AR can be used in education, support, services, and so on. AR applications is the ability to directly prove the value, and not just let the fun feeling more comfortable and more, you can unearth some vertical applications in some areas, this is a cognition.

China opportunities in AR and VR, I think may be associated with cock wire groups and value for users, you can observe for a period of time. The competition threshold is relatively low, but whether the rapid development, may be time.

5, Google's ambitions

Also went to Google to see the Sundar Pichai (GoogleCEO Pi Cai), Ms LI, a Professor of artificial intelligence at Stanford, as well as Hoffman, the founder of the British. And they talk about topics related to artificial intelligence, will not say one can share my conclusion.

Last year Google adjust the Alphabet, in fact, we know why they do it, but this time to a more profound understanding. Basically, Google wants to do a "machine head", this "brain" specifically, it is a combination of the following are a few things.

First, you need to have a particularly large amount of data, and the amount of data the best is not public, is your private, and are constantly updated, added. Because you do not have it, you have no competitive advantage.

Second, you have to have a particularly large machine platforms, capable of operation above, learning, iteration, your "brain" more and more intelligent, and this data can do better.

Third, you need to study a number of great depth or machine learning expert, they know how to get massive servers and huge amounts of data, turning data into knowledge and from the inside, and can do things.

Once you have these three things, you can apply to other areas. Used on a search, a search sort--this is "brain" can index all of the world, then you may search, I can tell you that one of the best sort. In the field of life, is a Google Now, and can tell you where to go for dinner today, you'd better take the subway to go to, then you can buy flowers on the way to remind your wife's birthday is coming up, is actually a combination of all these things up. In the field of advertising, an ads is how can you make more money.

Before and our IT people are always thinking of how to use IT to make life better, why do tend to ignore this data cannot be used in genome sequencing? Why can not be used in the field of biotechnology, pharmaceuticals, health? Or are all areas you can imagine, because once you have this huge data, your value is huge.

So as you can see, Google from Genentech (United States Genentech Corporation) dig CEO, to make Alphabet has a pharmaceutical company CEO, so Google's ambitions are clear. He used search and advertising to create a huge "brain", the "brains" of smart engineers to adjust, with huge data to study, plus the huge amount of computation to continue iterating. Then put this three who distribution to with, find a field of new data came in, like we to learn cancer of treatment, if can has a what library--a national of all of a gene, and cancer cases, then let data roll up, you also with hospital combined up, has a feedback of way, know whether effective, constantly to track, real-time Diego generation, may on will master cancer treatment of method.

So Google's directions, or direction of the Alphabet is constantly looking for new areas, to find a leader in the field, with related data, and several machine learning experts, give them a lot of machines used to calculate, you can generate value . This invincible in any field. Help young people to find the object, what they eat, speculate, of all things, even the military, there is no problem.

Boiled down, Alphabet ambition was to become a do everything, "brains" to drive and a subversion of the traditional industry companies. They must have a lot of internal methods to analysis, which of the following areas, medical, construction, real estate, finance or secondary market or something like that.

To cite an example, Alphabet to make a banking application, to analyze your credit and risk. You find Bank take 10 million, if only see Bank internal information, so Bank only know you in here save has 5 million, and in media work, information, but if I has addition a reptile, can put you of other data are climbed to, like you also in United States Goldman hid has 20 million, and in Cayman Islands buy has a building House, boom dub, "brain" on will told you can borrow to he.

So the Alphabet so that may become the greatest in the world, but also the most horrible company. Of course I think it is the ambition of the company a lot, but the Alphabet is most based company doing well.

This also raises some very important questions. The first question, there is such a large company, he should pay a social responsibility, self control? Don't be evil is a problem. Do not hurt humans or user is another question. So Google has set up an Ethics Committee to review his development of artificial intelligence.

Associated with this is, when after the Super artificial intelligence, it is really helping people, or will it destroy a man? The topic I had specialized in the CMU commencement speech, talking about our responsibility as a computer scientist.

李开复：我在硅谷看到了科技的未来 - 谷歌,Android,VR - IT资讯

15天，100人，2016年新年伊始，李开复亲自带队奔赴硅谷。

26位鼎鼎大佬，DST米尔纳、Google皮猜、雅虎杨致远、YC孵化器SAM、安卓之父Andy Rubin……

18家时下最富盛名企业，苹果、谷歌、Facebook、Airbnb、特斯拉、推特、YC孵化器……

科技创新圣地硅谷有哪些趋势正在发生？什么样的科技会在下一阶段席卷全球、改变世界？通过投资奠定了当下互联网格局的DST创始合伙人米尔纳如何找“百亿美元公司”创始人？安卓之父Andy Rubin正如何重复他的安卓奇迹？Google升级Alphabet背后怎样的雄心抱负？

近日，在中国“硅谷”中关村，李开复向媒体分享了他的新年“硅谷见闻”。这些见闻包括他对当前中国创业者的思考和建议、硅谷最前沿的科技、正在发生的科技趋势、结合谷歌的战略可以预见的未来，以及对于今年火热当中的VR和AR的思考。

以下为李开复口述实录的整理：

1、机器人

先讲一些比较吸引眼球的东西吧。我觉得在每个会议上都有一些有趣的讨论。比如跟Andy Rubin（Android之父）的讨论就非常有意思，他做的公司叫Playground Global，其实《华尔街日报》上已经报道了很多。见到他的那天，我们正好看到了那个Boston Dynamics（谷歌旗下军用大型机器人公司波士顿动力）的那只机器狗。

这只机器狗很好操作，我也玩了一下。说起来还挺有缘分的，因为Boston Dynamics的创始人Marc Raibert，之前是CMU（媒体注：卡内基梅隆大学，李开复母校）的教授，我是学生的时候，他的办公室就在我隔壁。那个时候他在CMU做的项目是一个会单脚跳，然后跳一分钟都不会摔倒的机器人。不过那个时候，如果你拿着棍子轻轻一碰，它就会倒了。而且当时还有一捆很粗很长的线，连接在电脑上，这就是最早的情形。

前几天刷屏的那个怎么踹都不会倒的机器狗和机器人，其实已经迭代进步了很多。当时Marc Raibert创立Boston Dynamics，并且获得了美国国防部的研究经费，专门做机器人研究。从一只脚做到四只脚，再做回两只脚，每一步推进都很不容易，所以已经花了美国国防部可能接近上亿美元的经费了。然后，谷歌看上了它就把Boston Dynamics买进来了，买进来以后谷歌就没有让它再拿国防部的钱。

当时买Marc Raibert的Boston Dynamics的决策者就是Andy Rubin。因为Andy Rubin做了Android之后，Google希望把Android和Chrome并起来，由Sundar Pichai（谷歌CEO皮猜）接替他的管理。然后创造了GoogleX，就把Andy Rubin请到Google X去里面做机器人，因为他自己爱好一直是机器人。而且Android本身也是手机机器人。我们和Andy Rubin交流那天，刚好Marc Raibert带着机器狗来拜访Andy Rubin，所以我们见到了。这是关于Google机器狗和机器人我所知道的。

2、机器人平台

Andy Rubin后来离开了Google X，创立了一个叫Playground Global的公司。简单来说，他的公司就是希望做一个机器人版的安卓平台。那我们如果以手机来参照的话，大概在十年前，你如果要开一个手机公司可能要花一两亿美金做研发，才能把手机做出来，但是现在你可能花一百万美金就能做出来了。因为你可以有各种代工，有标准模块，然后软件用安卓，还有其他。如果你不要什么特色，你就是要搞一个手机出来。那现在来说，创造手机这个的硬件的成本已经被降低了一百倍，然后就普及了。所以你看现在乐视为代表的互联网手机厂商全都跑出来了。当然，小米创业的时候做手机还是挺贵的，不过在此之后就越来越便宜。

同样的，Andy Rubin也认为说，如果机器人要普及这一天也必然发生，那他做的事情就是希望降低机器人创业模块之类门槛。比如安卓提供了智能手机的模块，让一个开发手机的，从手机硬件，一直到软件都变得容易，而且成本降低，让更多人进来了，这一下才能真正让更多人围绕手机进行创业，要不然创业门槛太高了。

Andy Rubin现在认为机器人也差不多在十年前智能手机状态下，所以做一个机器人平台，让更多的人来做机器人的创业，这会是一个改变世界的事情，也是他从孩童时代的一个梦想。所以Playground Global就是一个Android for robot，但是它不只是那种会走路的机器人。如果你要做一个简单的工业机器人，或者是一个像IFA的那个扫地的机器人，都是可以的。从机器人模块的角度来看，机器人基本就是一大堆传感器，组合起来，然后有学习训练和控制，让它能够动——动手、动脚、动它的爪子之类的。此后再有其他部分里加入系统、输入和识别之类。

这就是Andy Rubin跟我们分享的他的梦想。他的模式其实跟创新工场的初期非常相似，基本上是由一个孵化器来深度参与一些项目，然后把其中挖掘到的有价值的模块拿来标准化。接着再把好的项目拆分出去，作为独立的公司发展。

当然他也和我们讨论，在中国有没有机会来做一个Playground，他已经拿了一些中国的投资，媒体有报道。至于未来是否会和创新工场合作，我们现在还不方便说。总之，我们很佩服他的远见，但他要做的东西，因为加了硬件，复杂度会变大很多，这个可能会是一个挑战。

而且就我看来，他的这次创业，跟小米在一定程度上有点类似。不是说他做的产品类似，而是做这个东西需要有一种巨大的能量，需要有一个创始人，可以无限地吸入资金、然后投资，接着来打造别人认为很难的事情。所以在这方面，我觉得雷军跟他有很相似的地方。因为他也有大批鼎鼎有名的投资人，而且投入非常巨大，估值又巨高，又是做的一个特别长远、困难度很大、风险很高的设计。

另外一个很有趣的事情，也有一点讽刺的意味在里边。Andy Rubin现在的办公室在Fry’s Electronics，这是硅谷极客多年买零件的地方，和中关村差不多，不过Fry’s Electronics就是一个统一的大卖场，只此一家。比较早的时候，像我们这些极客到了周末就会去那边看，有什么便宜的PC板，然后可以买回家拼凑，或者家里的硬盘没有了，去那边看有没有打折，总之就是各种电子元器件的批发市场。

但是后来因为电子商务，大家买东西就不去卖场了，所以Fry’s electronics就卖掉了一半的楼层给Andy Rubin。他也跟我们说了下他的“野心”，他说Fry’s Electronics一定是要倒闭的，我现在就拿他楼层的一半，然后等着它倒闭，它倒一层我拿一层。

所以Andy Rubin还是蛮有意思的，有一些收获是：在最前沿的领域里面，在机器人和智能硬件的这个领域里，我们怎么去重复安卓的奇迹。

3、人工智能

另一个很有趣的现象，是做深度学习的人工智能博士生，现在一毕业就能拿到200到300万美金的年收入的offer，这是有史以来没有发生过的。当然我觉得硅谷的公司都在追捧这个方向，而且基本是四大名校：斯坦福、MIT、CMU、伯克利。以前这些学校的博士生在硅谷都可以拿到高薪，但是从来没有到250万美元/年的薪资水平。

这为什么会发生呢？第一，是因为真正懂深度学习的人现在还不是很多，所以供需不平衡。

第二，是因为很值。Google拿到这样的人，他就可以马上用他赚一百倍的钱。因为你只要把这样的一个人用在某个领域，比如说，假设谷歌要用他手上的现金做最聪明的二级市场的财务投资，这一个人一年就赚出一百倍来，所以这事毫无疑问是划算的。

第三，是因为涉及竞争。Google会很不希望这样的人落入他的竞争对手怀中。因为Google有这样一个领先的优势，但如果这个人去了Facebook、微软，马上就会给后两家机会。所以现在是一个关于人才的战争。对于这一批一年可能少于50个的博士毕业生，这三大公司：Google、Facebook和微软，都在用不合理的价钱去挖。

这个给了我两个启示。一方面是遗憾自己生太早了，我就在这个领域，但是那时候没有公司这样来挖我们。另一方面是类似的人才战争，可能也会在中国发生。

中国的大学恐怕没有这样的50个博士，但是我们这边有几所“大学”里毕业的，也是跟硅谷一样的。这几所“大学”的名字叫做腾讯、百度和阿里巴巴。所以这些人未来会不会因为数据的诱惑，或者对公司的忠诚留下来做点事情呢？还是这几家公司之间互挖，可能会成为很有趣的事。

因为我自己是做这方面的，所以我觉得深度学习虽然很厉害，但是没那么了不起。你让一个聪明的人学两年，他也可以有这个价值。所以这也就是说，我们是不是应该来帮助培训一千个中国的深度学习专家，这些都是很有趣的讨论。

然后我也问了他们，你们这样拼命的竞争，再过两三年，中国学这些东西也不难，中国数据也比你们多，你们这套公开了，中国在这个领域的人才可能会比美国多。

因为在中国，百度、腾讯、新浪微博等等的数据量也非常大，如果在中国也有250万美金的年薪诱惑，中国人会更疯狂地冲向这个领域。所以我相信这个领域最后应该也是一个中美领跑的状态，虽然现在中国是远远落后美国，毕竟美国有斯坦福、伯克利，CMU、MIT出来的人，而且是不断流动。但对于中国来说，这批人可能就在百度、腾讯和阿里。现在百度、腾讯和阿里自己hold得住这些人，但是长期来说他们也不可能永远hold住。而且长期来说，小米、奇虎360也都会有这样的人，所以这会是一个很有趣的业界竞争的状态。

于此相关的，领英的霍夫曼提到马斯克和彼得-蒂尔他们成立了一个开源平台Open AI。这个Open AI成立主要是为了防止Google、Facebook和微软这样的大公司形成垄断，你们有那么多计算机，那么有钱，又把最优秀的人全挖进去了，所以我们就要搞一个开放式的开源平台，来确保这个东西能够被更多的人快速学习掌握掉。所以这是一个很有趣的事情，硅谷考虑得很超前。

4、VR和AR

整体来说，关于VR各AR，主要感受主要有这么几点。

先那VR来说，对于这个领域的发展，有一批人是非常乐观的，另外一批人则认为我们还早了一个cycle——就是说现在还在摸索状态，因为内容不够多、体验不够好、太贵，然后还要连一个PC，拉一根线，可能真的还在一个玩家的时代。

就创新工场来说，我们也综合了一些意见。我们的看法是，VR长远来讲对社会的影响应该是特别巨大的，而且随着摩尔定律等作用，它应该会越做越炫，越做越不头昏，越做越没有线，越做越轻，越做越小。最终变成一个，也许不是眼镜，但是是某一种模式，让你不知不觉就把它融入到你的生活里面去了，这一天是绝对会到来的，会是一个巨大的产业，会改变所有的事情。

但是具体方面，我们可能会稍微保守一点。我们可能对这种五到十年的未来，抱有一个很乐观的期待。但是对于他到底能不能在一两年之内，Oculus也好，HTC的产品也好，他们在一两年之内能不能打破一个玩家的领域，能不能达到普及的状态，我们可能还是抱一个观看和怀疑的态度。

当然我觉得从投资的角度，现在看到好公司就得投了，因为你不能进入cycle才投，只是说我们如果对它期望值很高，说它要颠覆什么的话，我们还有一点时间。

还有就是从应用层面来说，VR我自己以前在SGI做过，做3D的东西。所以我个人对于VR的看法是，它的第一个突破一定是在娱乐方面。因为我们讲了那么多3D的东西，都还没有被验证，而且人们对3D的需求基本上只有在娱乐内容相关的领域得到了验证，所以基本上就是看电影更爽、玩游戏更爽、然后越做越逼真，大概是这样一个状态。有点像电影业和游戏业的一个延伸，但需要说明的是，这个延伸是一个巨大的延伸。

AR来讲，它是可以有不同领域的应用。AR可以用在教育方面、辅助方面、服务方面等等。AR的应用是能够直接被证明价值的，而不只是让娱乐感更爽更强，可以在一些领域挖掘出一些垂直性的应用，这是大家的一个达成的认知。

中国在AR和VR方面的机会，我觉得可能会跟屌丝群体和性价比用户相关，可以再观察一段时间。这个领域竞争的门槛会相对比较低，但到底能不能快速发展，可能还要一点时间。

5、谷歌的野心

此外还去了Google见了Sundar Pichai（GoogleCEO皮猜），斯坦福人工智能方面的教授李菲菲，以及领英的创始人霍夫曼。与他们谈论的主题都是人工智能相关的，具体的就不一一说了，可以把我的总结分享一下。

去年Google调整成Alphabet，其实我们也知道他们为什么这么做，但这次去了就更加深刻地了解了。基本上，Google想要做一个“机器大脑”出来，这个“大脑”具体来讲，它其实是下列几件事情的结合体。

第一，你要有特别大的数据量，而且这个数据量最好不是公开的，是你私有的，而且是可以不断地更新、增加的。因为你要没有这个东西，你就没有竞争优势。

第二，你要有特别巨大的机器平台，能够在上面运作、学习、迭代，让你的“大脑”越来越聪明，而且用这个数据能越做越好。

第三，你需要一批特别棒的深度学习或者机器学习的专家，他们知道怎么去弄海量的服务器和海量的数据，从里面把数据变成一种认知和知识，以及能做的事情。

一旦有这三件东西之后，你是可以应用到其他领域的。用在搜索上，就是一个搜索排序——这个“大脑”能把世界全部索引了，然后你搜什么我就能告诉你，做一个最好的排序。用在生活领域，就是一个Google Now，可以告诉你今天要去什么地方吃饭，你最好搭地铁去，然后路上可以买花，提醒你老婆生日快到了，实际上是把这些东西都结合起来了。用在广告领域，就是怎样投放一个广告能让你赚更多钱。

而且之前我们这些IT人总是想着如何用IT让生活更美好，往往忽略了这套数据为什么不能用在基因排序？为什么不可以用在生物科技、制药、健康领域？或者是你可以想象的所有领域，因为一旦你有了这个巨大的数据，你的价值就巨大了。

于是你可以看到，Google从Genentech（美国基因泰克公司）挖了CEO，来做Alphabet里面的医药公司的CEO，所以Google的野心是非常清晰的。他用搜索和广告来塑造了一个巨大的“大脑”，这个“大脑”让聪明的工程师来调整，用巨大的数据来学习，加上巨大的计算量来不断地迭代。然后把这三者配到一起，找一个领域的新数据进来，比如说我们要学癌症的治疗，假如能够有一个什么库——某个国家的所有人的一种基因、癌症病例，然后让数据滚起来，你还跟医院结合起来，有一个回馈的途径，知道是否有效，不断去追踪，实时迭代，可能就会掌握癌症治疗的方法。

所以Google的方向，或者Alphabet的方向就是不断地找新领域，找一个该领域内的领军人物，拥有相关的大数据，再配几个机器学习专家，给他们一大堆机器用来计算，就能产生价值了。这会在任何领域都攻无不克。帮年轻人找对象、吃什么、推测所有的事情，甚至军事，都没有问题。

归结起来，Alphabet的野心就是成为一个无所不为，用“大脑”来驱动并颠覆传统行业的一个公司。他们一定有很多内部的方法来分析，接下来开展哪个领域，是医学、建筑、房地产、金融还是二级市场之类的。

举个例子，Alphabet做一个银行相关的应用，来分析你的信用和风险能力。你找银行借一千万，如果只看银行内部资料，那么银行只知道你在这里存了五百万、在媒体上班等信息，但如果我有另外一个爬虫，能把你的其他数据都爬来，比如你还在美国高盛藏了两千万、在开曼群岛买了一栋房子，咚咚咚，“大脑”就会告诉你可以借钱给他。

所以Alphabet这么一来，可能就会成为世界上最伟大，同时也是最可怕的公司。当然我觉得有这个野心的公司其实很多，但是Alphabet应该是最有基础把它做好的公司。

于是这也就引发了几个很重要的问题。第一个问题，有这么大数据量的公司，他应该如何付出社会责任，实现自我管制？不作恶是一个问题。不伤害人类、伤害用户是另一个问题。所以Google设立了一个道德委员会专门用来审核他在人工智能方面的一些发展。

与此相关的是，当这个超级人工智能出来以后，它是真的帮助人，还是会毁灭人的？这个话题我在CMU的毕业典礼上专门做过演讲，谈到我们作为计算机科学家的责任感。