摘要:目前为止,我们已经对硅谷的各项新技术:人工智能(A.I.)和聊天机器人耳熟能详了。这些新技术给我们一种希望,那就是它们能够带来像个人电脑和电子商务一样的大规模改变。

目前为止,我们已经对硅谷的各项新技术:人工智能(A.I.)和聊天机器人耳熟能详了。这些新技术给我们一种希望,那就是它们能够带来像个人电脑和电子商务一样的大规模改变。

针对研究人工智能的公司的投资额近年来增加了四倍不止。根据CB Insight公司市场调查的数据,2015年已经达到了681,000,000美元,而2011年仅仅有145,000,000美元。CB Insight预估近年投资额将会增加到1,200,000,000美元,比去年增加76%。

显然大家的期望值很高,但是和别的科技一样,A.I.也有很多限制。而在这个领域,技术恰恰是最大的限制。第一个A.I.平台花费了60年才通过图灵测试。最起码我们还要等另一个六十年,A.I.才有可能离开人类的帮助独立工作,何况这一天有可能永远不会到来。但目前为止,我们可以看看我们在哪些领域遇到了障碍。

 

  1. 我们还是需要一些人情味

Fackbook’的 M平台 是首个证明为了完善A.I. 人类是不可或缺的一部分的智能系统。这个平台将机器和人工客服结合起来,并且取得了成功,而与此同时,其他很多像Siri一样的智能聊天系统的成长之路却并不通顺。一个名为“trainers”的客户服务中介,监控了M平台的每一次聊天过程,所以M平台看起来这么智能也就不奇怪了。然而,M平台的成功也引发了一个新的问题,如果真人是M平台发展成功的最重要的原因的话,为什么还要用A.I呢?M平台真的算是应用A.I.技术取得成功的例子吗?

还有就是关于人类感知能力的问题。Siri这样的机器并没有任何自我感知力。如果你想要用Siri帮你订机票,你是不能指望它有自己的判断的。如果你只想订一个最便宜的机票,人工智能助手不会告诉你,那个只便宜十块钱的航班要多飞10小时。最后,你的确坐到了最便宜的航班,同时你也会因此而头疼不已。

机器都是按照它们的编程去运作的,所以它们自然也不会像人类一样可以随机应变。

除此之外,机器还不能理解一些诸如同情心,创造力,理解力和合作这样的概念。无论它们多么聪明,机器永远都会缺少人情味。

 

  1. 机器不会夺走我们的工作

最成功的A.I.企业是那些利用知识和A.I.来优化雇员的企业,而不是那些想着利用A.替换掉雇员们的企业。

最聪明的企业的确会以A.I.的形式去支撑起一个有更多权利和资源的核心小组。然而这仍不能解决所有的问题。再重申一遍,人类还是至关重要的,否则谁来给命令,不可能是机器吧。谁来出主意,也不会是机器。

 

  1. 机器人不会统治世界

如果能够制造出完全模仿人类行为和思考方式的机器,那的确是一项了不起的科学成就。但是即使是现在的A.I.领域的顶级专家也无法担保真的能发明出来这样的机器。所以,尽管现在人工智能不断发展,有些时候A.I.仍然不会成为我们的首选。

以前我们使用A.I.来预订会议,而现在我们可以用A.I.去订机票和下菜单。A.I.的功能总是为我们减少一些日常琐事。但是绝大多数的像Siri一样的A.I.平台,总是理解错误,反而给我们带来了更多的工作。

苹果的IOS 9大大的提升了Siri的沟通能力,但是A.I.目前为止还是让人非常失望。所以无论A.I.发展到什么程度,人类还是占据统治地位的。

 

(文章来源:http://venturebeat.com/2016/07/30/why-chatbots-will-never-be-truly-useful/  原文作者:BARNABY LASHBROOKE, TIME ETC.  鲸媒体编译。)

 

原文:

Why A.I. will never be truly useful

 

By now, we’ve all heard of Silicon Valley’s shiny new technology: Artificial intelligence and chatbots. They bring the promise of an industry-wide transformation that many believe will reach the scale of the personal computing industry or the commercial internet.

Funding in artificial intelligence startups has increased more than fourfold — to $681 million in 2015, from $145 million in 2011 — according to the market research firm CB Insights. The firm estimates that new investments will reach $1.2 billion this year, up 76 percent from last year.

It’s clear that hopes are high, but, as with any technology, there are limitations. In this case, the technology is the limitation. It took six decades for the first A.I. platform to pass the Turing Test, and we are most likely at least another six decades away from machines being able to ever act without the help of humans — if, indeed, that day ever comes. But, for now, let’s take a look at some of the areas where the hiccups are occuring.

 

  1. We still need the human touch

Facebook’s M is the first A.I. platform to prove that humans may have been the missing ingredient in previous efforts to perfect A.I. With its mix of machine-based and human customer service, M is succeeding where many voice assistants, like Siri, have stumbled. A team of customer service agents — called “trainers” — actually monitor every M communication. No wonder the platform seems so intelligent. However, this begs the question, if humans are what’s so great about M and are essential to the platform, why even use A.I. to begin with? And is M an example of true A.I.?

Then there’s the matter of human perception. Machines like Siri are not intuitive, and they can’t use their “judgment” when it comes to something like booking a flight. If you simply ask for the cheapest flight, an artificial assistant won’t take the initiative to tell you that one flight is slightly cheaper than another but also 10 hours longer. You’ll get the cheapest flight, but you’ll also get a headache.

Machines cannot deviate from what they are programmed to do, and they are definitely not as good at altering their responses to changing situations as humans are.

Beyond that, machines simply cannot understand concepts such as caring, creativity, understanding, and togetherness. No matter their level of intelligence, machines will always lack the human touch.

 

  1. Jobs are safe

Innovative companies that take on A.I. successfully will be those that combine human knowledge with A.I. in order to optimize their employees, not replace them.

Smart companies are buttressing a core group of employees with more power and resources in the form of A.I. This helps with some tasks but still not with all. Again, humans are needed. Who’s giving the direction? Not a machine. Who’s coming up with ideas? Not a machine.

 

  1. Bots will never take the helm

The ability to create machine intelligence that mimics human behavior and thinking would be a huge scientific accomplishment, but even A.I. experts are not promising it will ever happen. So, while smart machines and A.I. continue to grow, there are some instances where they will never be the preferred method.

The days of using A.I. to book meetings has given way to A.I. booking flights and food orders. The goal of A.I. has always been to minimize the pain of mundane tasks, but most A.I. platforms, like Siri, actually end up causing the user more work, due to numerous mistakes and miscommunication on the part of the machine.

Siri’s communication has definitely improved with Apple’s iOS 9 update, but A.I. has actually been a huge letdown, so far. And no matter how far it comes, humans will always be at the helm.