lzy 于2017-10-20发布 l 已有人浏览
增大字体 减小字体


I’m here to offer you a new way to think about my field, 今天来到这里,是想邀请大家从一个新的角度来看我熟悉的领域:
artificial intelligence. 人工智慧。
I think the purpose of AI 我认为人工智慧的目的,
is to empower humans with machine intelligence. 是要用机器智慧来增强人类的能力。
And as machines get smarter, 而随着机器变得更聪明,
we get smarter. 我们也会更聪明。
I call this "humanistic AI" -- 我称这个为「人性化人工智慧」──
artificial intelligence designed to meet human needs 透过与人工智慧合作并强化人类的设计方式
by collaborating and augmenting people. 来满足人类的需求。
Now, today I’m happy to see 如今,我很高兴能看到
that the idea of an intelligent assistant 智慧助理的这个想法
is mainstream. 能成为主流。
It’s the well-accepted metaphor for the interface between humans and AI. 这象征了人类与人工智慧之间的互动介面已被广泛接受。
And the one I helped create is called Siri. 而其中一个我协助创造的就是 Siri。
You know Siri. 各位都认识 Siri。
Siri is the thing that knows your intent Siri 会知道你的意图,
and helps you do it for you, 并协助你
helps you get things done. 把一切搞定。
But what you might not know is that we designed Siri 但各位可能不知道,当初我们在设计 Siri 时,
as humanistic AI, 就把它定位在「人性化的人工智慧」,
to augment people with a conversational interface 透过对话介面的方式来强化人类,
that made it possible for them to use mobile computing, 让使用行动计算成为可能,
regardless of who they were and their abilities. 不论他们是谁、能力如何,都可以使用。
Now for most of us, 对我们大多数人而言,
the impact of this technology 受到这项技术的影响
is to make things a little bit easier to use. 是让我们在使用东西上变得比较容易。
But for my friend Daniel, 但对我朋友丹尼尔而言,
the impact of the AI in these systems is a life changer. 却是改变了他一生的人工智慧系统。
You see, Daniel is a really social guy, 丹尼尔是个很爱社交的人,
and he’s blind and quadriplegic, 他也是个四肢麻痺的盲人,
which makes it hard to use those devices that we all take for granted. 我们认为理所当然的装置设备,对他来说却很难用。
The last time I was at his house,his brother said, 上次我去他家,他哥哥说:
Hang on a second, Daniel’s not ready. 「等等,丹尼爾還沒準備好。
He’s on the phone with a woman he met online." 他正在和他线上认识的女生讲电话。」
I’m like, "That’s cool, how’d he do it?" 我说:「好酷,他怎么办到的?」
Well, Daniel uses Siri to manage his own social life -- 丹尼尔用 Siri 来管理他自己的社交生活──
his email, text and phone -- 他的电子邮件、讯息、电话──
without depending on his caregivers. 不用依赖他的照护者。
This is kind of interesting, right? 这挺有趣的,对吧?
The irony here is great. 这也相当讽刺。
Here’s the man whose relationship with AI 这个人与人工智慧的关系
helps him have relationships with genuine human beings. 协助他与真正的人类建立了关系。
And this is humanistic AI. 这就是人性化的人工智慧。
Another example with life-changing consequences 还有一个可以改变人生结果的例子,
is diagnosing cancer. 也就是癌症诊断。
When a doctor suspects cancer, 当医生怀疑你有癌症时,
they take a sample and send it to a pathologist, 他们会先取样交给病理学家,
who looks at it under a microscope. 病理学家再利用显微镜来看样本。
Now, pathologists look at hundreds of slides 病理学家每天要看数百片的载玻片
and millions of cells every day. 及数百万个细胞。
So to support this task, 所以为了支援这项工作,
some researchers made an AI classifier. 有几位研究人员做出了一种人工智慧分类器。
Now, the classifier says,"Is this cancer or is this not cancer?" 现在,分类器已经会看着图片说:
looking at the pictures. 「这是不是癌症?」
The classifier was pretty good, 分类器挺厉害的,
but not as good as the person, 但没有人类那么厉害,
who got it right most of the time. 人类大部份都能判读正确。
But when they combine the ability of the machine and the human together, 但,当机器与人类的能力结合在一起时,
accuracy went to 99.5 percent. 正确率可以达到 99.5%。
Adding that AI to a partnership eliminated 85 percent of the errors 加上人工智慧的协作,人类病理学家
that the human pathologist would have made working alone. 能减少个人判断时 85% 的错误。
That’s a lot of cancer that would have otherwise gone untreated. 这相当多啊,很多癌症患者,当初就是因误判而耽搁了治疗。
Now, for the curious, it turns out 人类因为好奇的原因,所以
that the human was better at rejecting false positives, 比较擅长判断错误的假阳性案例,
and the machine was better at recognizing those hard-to-spot cases. 而机器比较擅长判断难以辨别的案例。
But the lesson here isn’t about which agent is better 但我们学到的并不是
at this image-classification task. 在比谁在图像分类上比较厉害。
Those things are changing every day. 那些东西每天都会变。
The lesson here 这里我们学到的是,
is that by combining the abilities of the human and machine, 借由结合人类与机器的能力,
it created a partnership that had superhuman performance. 其创造出的合作关系会有超人般的表现。
And that is humanistic AI. 而这就是人性化的人工智慧 。
Now let’s look at another example 我们再举个例子,
with turbocharging performance. 这方面的进步也是突飞猛进。
This is design. 那就是「设计。」
Now, let’s say you’re an engineer. 假设你是个工程师,
You want to design a new frame for a drone. 你想要设计一个新的无人机骨架。
You get out your favorite software tools, CAD tools, 你拿出你最爱的软体工具,电脑绘图辅助工具,
and you enter the form and the materials,and then you analyze performance. 你输入了外形和材料,然后按下分析表现。
That gives you one design. 它只会跑出一种设计方案。
If you give those same tools to an AI, 如果你把这些工具拿给人工智慧使用,
it can generate thousands of designs. 它会帮你跑出数千种设计。
This video by Autodesk is amazing. 欧特克做的影片非常震撼。
This is real stuff. 这是真实案例。
So this transforms how we do design. 这改变了我们设计的方式。
The human engineer now 现在,人类工程师
says what the design should achieve, 只要说出想要达成的设计,
and the machine says, 机器就会说:
Here’s the possibilities. 「有这些可能性方案。」
Now in her job, the engineer’s job 现在,她的工作,工程师的工作,
is to pick the one that best meets the goals of the design, 就是选出一种最符合目标的设计,
which she knows as a humanbetter than anyone else, 这是她身为人类最擅长的,
using human judgment and expertise. 用的就是人类的判断和专业。
In this case, the winning form 在这案例中,脱颖而出的设计
looks kind of like something nature would have designed, 像是自然界设计出来的,
minus a few million years of evolution 但这样的设计,不需要几百万年的演化
and all that unnecessary fur. 也不需有那些皮毛。
Now let’s see where this idea of humanistic AI might lead us 我们现在来想想,如果我们跟着人性化人工智慧走,
if we follow it into the speculative beyond. 它会把我们带到那些意想不到的境界?
What’s a kind of augmentation that we would all like to have? 我们想要的强化,有哪些方面?
Well, how about cognitive enhancement? 认知强化如何?
Instead of asking, 别再问:
How smart can we make our machines? 「我们能把机器做到多聪明?」
let’s ask 我们要改问:
How smart can our machines make us? 「我们的机器能把我们变多聪明?」
I mean, take memory for example. 我的意思是,例如「记忆力。」
Memory is the foundation of human intelligence. 记忆是人类智慧的基础。
But human memory is famously flawed. 但人类的记忆力是出名的差。
We’re great at telling stories, 我们很会说故事,
but not getting the details right. 但无法把细节说正确。
And our memories -- they decay over time. 我们的记忆力会随时间衰退。
I mean, like, where did the ’60s go, and can I go there, too? 就像六十年代去哪了?我也能去那儿吗?
(Laughter) (笑声)
But what if you could have a memory that was as good as computer memory, 但如果你能拥有和电脑一样好的记忆力,
and was about your life? 而且这些记忆与你的人生有关,那会如何呢?
What if you could remember every person you ever met, 如果你能记得你见过的每一个人、
how to pronounce their name, 叫得出每一个人的名字、
their family details, their favorite sports, 家庭状况的细节、他们最爱的运动、
the last conversation you had with them? 你和他们上次的对话,那会如何呢?
If you had this memory all your life, 如果你一生都有这样的记忆,
you could have the AI look at all the interactions 你就可以要求人工智慧去看看你过去
you had with people over time 与人们的所有互动,
and help you reflect on the long arc of your relationships. 并协助你反思你的人际关系弧线。
What if you could have the AI read everything you’ve ever read 如果我们能让人工智慧去阅读所有你阅读过的东西、
and listen to every song you’ve ever heard? 去听你听过的每首歌,会如何呢?
From the tiniest clue,it could help you retrieve 从最微小的线索,它就能协助你找回
anything you’ve ever seen or heard before. 你以前看过或听过的任何东西。
Imagine what that would do for the ability to make new connections 想像一样,那对于
and form new ideas. 建立新连结以及产生新点子的能力会有什么影响。
And what about our bodies? 那么,我们的身体呢?
What if we could remember the consequences of every food we eat, 如果我们能记得我们吃过的每样食物、
every pill we take, 每颗药物、
every all-nighter we pull? 记得每次熬夜 K 书的内容,那会如何?
We could do our own science on our own data 我们可以用自己的资料来做自己的科学,
about what makes us feel good and stay healthy. 找出什么会让我们感觉很好并保持健康。
And imagine how this could revolutionize 想像一下,这会为我们管理
the way we manage allergies and chronic disease. 过敏及慢性病带来怎样的革命。
I believe that AI will make personal memory enhancement a reality. 我相信,人工智慧能够实现个人记忆的强化。
I can’t say when or what form factors are involved, 我不知道何时或需要那些机缘,
but I think it’s inevitable, 但我认为它是一定会到来的,
because the very things that make AI successful today -- 因为人工智慧在现今能够成功的原因,
the availability of comprehensive data 就是这些综合性资料的可取得性,
and the ability for machines to make sense of that data -- 以及机器能从这些资料中理解出意义的能力,
can be applied to the data of our lives. 能为我们生活上的资料所用。
And those data are here today,available for all of us, 我们现今就已经能够取得那些资料了,
because we lead digitally mediated lives, 因为我们可以直接用手机或上网的方式,
in mobile and online. 将我们的生活与数位科技做连结。
In my view, a personal memory is a private memory. 依我所见,个人的记忆是私人的记忆。
We get to choose what is and is not recalled and retained. 我们可以选择要或不要,回忆或保留哪些记忆。
It’s absolutely essential that this be kept very secure. 保持资料的安全绝对是有必要的。
Now for most of us, 对大部份人来说,
the impact of augmented personal memory 增强个人记忆希望产生的影响会是:
will be a more improved mental gain, 可以获得心智上的成长
maybe, hopefully, a bit more social grace. 或是希望可以有更高雅的社交能力。
But for the millions who suffer from Alzheimer’s and dementia, 但对于数百万受阿兹海默症及痴呆症所苦的人而言,
the difference that augmented memory could make 增加记忆能造成的改变,
is a difference between a life of isolation 就是让那孤独的人生,
and a life of dignity and connection. 变成有尊严及连结的人生。
We are in the middle of a renaissance in artificial intelligence right now. 我们现在正在人工智慧的文艺复兴时期当中,
I mean, in just the past few years, 仅仅在过去几年,
we’re beginning to see solutions to AI problems 我们就开始看到人工智慧问题的解决方法,
that we have struggled with literally for decades: 这在过去数十年来,一直是无法解决的问题,像是:
speech understanding, text understanding, 口语理解、文字理解、
image understanding. 图像理解。
We have a choice in how we use this powerful technology. 我们可以选择要如何使用这强大的技术。
We can choose to use AI to automate and compete with us, 我们可以选择用人工智慧来做自动化并与我们竞争、
or we can use AI to augment and collaborate with us, 或我们可以选择用人工智慧来做增强化并与我们合作,
to overcome our cognitive limitations 来克服我们的认知限制,
and to help us do what we want to do, 并协助我们去做我们想要做的事,
only better. 而且做得更好。
And as we discover new ways to give machines intelligence, 随着我们发掘出新的方式来将智慧赋予机器,
we can distribute that intelligence 我们可以把那智慧移转到
to all of the AI assistants in the world, 全世界所有的人工智慧助理身上,
and therefore to every person,regardless of circumstance. 且不论如何,都要确保每个人都能因此受惠。
And that is why, 那就是为什么
every time a machine gets smarter, 每次当机器变更聪明,
we get smarter. 我们也会变更聪明。
That is an AI worth spreading. 而这就是人工智慧值得推广的原因。
Thank you. 谢谢大家。
(Applause) (掌声)


 1 2 下一页