Can AI Make The World Happier and More Peaceful

thank you thank you

you may have heard of a story wherein

artificial intelligence is trying to

defeat

the human race and take over the world

or maybe seen a movie where a man

falls in love with an ai girlfriend

because his real girlfriend left him

i found it interesting that we so often

hear

about artificial intelligence in a

fearful or

funny context we don’t even realize it

but from medical devices to smartphones

air is already a part of our lives

at its core ai is a fundamental tool

that can help solve our most pressing

problems

in fact all the unjustified fear around

ai

is preventing us from properly using ai

to make our world a better place

it is unfortunate that despite

tremendous advances in technology

we still see millions of people going to

bed hungry

natural calamities like earthquakes and

floods

shatter millions of lives every year

can artificial intelligence help solve

these serious problems that we as humans

are facing every day

the answer is yes in this

talk i want to tell you how ai will make

the world

happier and peaceful hello everyone

i am pranav an engineering lead at

google research

i work on building artificial

intelligence to provide magical product

experiences for our users

recently i’ve been using ai to help

solve difficult humanitarian problems

that our global society has been facing

since a very long time

we all have probably heard the term ai

at some point or the other nevertheless

it would help to introduce it quickly

before we dive deeper well

artificial intelligence is the science

where machines can

self-learn from the data to make

themselves smarter

so instead of us giving it instructions

ai machine can go through patterns of

data

and learn to do tasks on their own

this is very similar to how we as

children learn to talk

walk or do things isn’t that fascinating

one important ai technology is neural

networks

neural networks are inspired by the

human brain

and work like the human brain just like

how our brain is organized as

nerve as layers of nerve cells called

neurons

neural networks are also structured in

the same manner

a small child observes their

surroundings

and learns things by observing patterns

around them

similarly neural networks learn by

observing many

examples and learning the connections

between the neurons based on those

examples

let us use a simple example to

understand this

we have an image at hand and the task

for our neural network

is to identify whether that image is

that of a cat or a dock

the way we solve it is by taking the

pixels of this image

and feeding it to the lowest layer also

called as the input layer

this layer processes this information

and then passes it on to the next higher

layer

this information flow is repeated

through multiple layers

until it reaches the top layer which has

two neurons

one corresponding to a dog and the other

corresponding to a cat

the neuron that gets more activated at

the top level

is essentially what gets predicted as

the final answer by this neural network

so this is how we use this neural

network to get

answers now one might wonder

how do we teach or train this neural

network

to teach it we feed in a bunch of

known examples examples where we already

know the answer

through this neural network when the

network gives

a wrong answer we use that opportunity

to back propagate the information

through the network

backpropagating is essentially a

feedback mechanism

using which we correct the edge weights

between the layers

such that the next time it is more

likely to give us a correct answer

this is just like learning from your

mistakes

on the other hand when the neural

network gives a correct answer

we reward it and we reinforce its ways

through the same process

very similar to how we reward a child

when they do the right thing

technique of back propagation learning

from your own mistakes

is what makes neural networks so

powerful enough to learn

almost anything

so now that we have an idea about the

basic concept

of ai and neural networks let us talk

about how it will help our society be

happier

air is already used in applications all

around us

ai is helping make our roads safer

through self-driving cars and smart

traffic signals

when you watch a video online ai often

helps you decide which video to watch

based upon your preferences

when you send a text message ai can even

help you draft that text message

through these applications ei is making

us more productive

by automating the mundane tasks and

happier by

freeing up our time to do things that we

love the most

[Music]

but some of the most challenging

problems that we continue to face

in the 21st century are problems like

hunger climate change or even illiteracy

given that we have ai in our quiver

we should expect and even rightfully

demand that it helps us solve these

difficult problems

let’s first talk about hunger about 690

million people

still go to bed on an empty stomach each

night

we are in the 21st century we claim

ourselves to be a modern civilization

so it’s simply not acceptable that so

many people suffer from lack of food

a major cause of hunger tends to be

calamities

not only natural calamities like floods

and earthquakes but also calamities

inflicted by humans like war

when a calamity strikes large

populations are cut off from basic

necessities like food and shelter

the global humanitarian community is

usually overwhelmed in the aftermath of

such a disaster

distribution of relief often tends to be

an even bigger problem than the

availability of relief material

so in partnership with the united

nations

we developed ai technology that can help

deploy aid

to disaster struck regions by analyzing

their aerial imagery

we analyzed that aerial imagery to

pinpoint the locations

that are most affected by damage and

thus

in critical need of aid

we solve this problem by using deep

learning

to compare before and after disaster

images of a region

when we compare the two images we were

able to determine where the most damage

had occurred as a result of the disaster

we do this using a two-step process the

first step

is to detect all of the buildings in a

given region

once we determine the buildings the

second step

is to determine which of these buildings

are damaged by the disaster

here’s an example from the haiti

earthquake

on the left side we can see buildings

identified as damaged

and on the right side we can see

buildings identified as not damaged

we can see that this technology is able

to make predictions that are at power or

even better than those made by humans

such information when aggregated

together

highlights regions that are hurt by the

disaster

this data this technology is hugely

useful to deploy

aid to people who are in critical need

of this aid

like food and medicine

this was an example of how ai is helping

solve

hunger but the potential of ai

extends far beyond to areas like health

care

and education in healthcare

i strongly believe that ai can make high

quality diagnosis accessible to the

entire planet

diagnosis that was previously accessible

only to the elite few

will become accessible to every single

person

we are already seeing this happen in the

field of diabetic retinopathy

which unfortunately has been a leading

cause of blindness

it’s truly saddening that most of these

patients did not have access to proper

diagnosis

as a result they didn’t receive adequate

treatment

and their health declined

using ai we now have diagnosis

technology

that performs at least as well as

doctors do

this is making diagnosis a lot more

scalable and accessible

this application is a perfect example of

human ai partnership

where an ai does the heavy lifting and

frees up our doctors to

spend more time interacting with and

helping their patients

i envision ai to work along similar

lines of human ai partnership

where both humans and ai work together

education education is the bedrock of

every society

after all that is what helps grow our

children to become capable citizens

unfortunately there are millions of

children out there

still without access to formal education

what if ei could make education more

accessible by providing a personalized

learning experience

to every child based on their strengths

and needs

every child is unique and deserves

access to high quality education that is

tailored

for them every field that we look

at be it agriculture or education or

healthcare can deeply benefit from ai

imagine a smartphone helping a farmer

to plan and plant crops get harvest

information

forecast weather forecast rainfall and

even get access to agricultural supplies

what is the use of technology if it

cannot truly uplift the repressed and

enrich their lives

friends ai is not science fiction

it’s a powerful tool that’s already here

to make our lives easier and happier

before closing i’ll say this

we cannot let irrational fear inhibit us

from using ai for human good

just like the steam engine invented in

the 18th century

or the telephone inventor in the 19th

century

ai has the potential to significantly

improve our lives

ei will help uplift millions of people

out of poverty

but just like the steam engine and like

the telephone

it will only be as useful as we make it

to be

so it is up to each of us to identify

such productive opportunities

and utilize ai to help heal the world

let us work together to use ai to make

our world

happier and more peaceful thank you very

[Applause]

much

you

谢谢谢谢

你可能听说过一个故事,其中

人工智能试图

击败人类并接管世界,

或者可能看过一部电影,其中一个男人

爱上了一个 AI 女朋友,

因为他真正的女朋友离开了他

我找到了 有趣的是,我们经常

在一个

可怕或

有趣的环境中听到人工智能,我们甚至没有意识到这一点,

但从医疗设备到智能手机,

空气已经成为我们生活

的核心,人工智能是一种基本工具

,可以帮助解决我们最 紧迫的

问题 事实上,围绕人工智能的所有不合理的恐惧

正在阻止我们正确使用人工智能

来让我们的世界变得更

美好 不幸的是,尽管

技术取得了巨大的进步,

我们仍然看到数百万人

饿着肚子上床睡觉

地震和洪水等自然灾害

破碎 每年数以百万计的生命

可以人工智能帮助解决

我们人类每天面临的这些严重问题

答案是 是的,在这次

演讲中,我想告诉你人工智能将如何

让世界

更快乐、更和平大家好,

我是 pranav,谷歌研究部的工程主管,

我致力于构建

人工智能,为我们的用户提供神奇的产品

体验

最近我一直在使用人工智能 为了帮助

解决

我们的全球社会长期以来一直面临的困难人道主义问题,

我们都可能

在某个时候或其他时候听说过人工智能这个词,但

在我们深入研究之前快速引入它会有所帮助

人工智能是科学

机器可以

从数据中自我学习以使

自己变得更聪明,

因此

机器可以通过数据模式

并学习自己完成任务,而不是我们给它指令,

这与我们

小时候学习说话

走路或 做事不是那么令人着迷

一项重要的人工智能技术是神经

网络

神经网络受到

人脑的启发

,工作起来像 人类的大脑就像

我们的大脑是如何组织成

神经一样的神经细胞层称为

神经元

神经网络的结构也

小孩观察

周围环境

并通过观察周围的模式来学习事物的方式

相同,神经网络通过

观察许多

示例来学习

根据这些例子学习神经元之间的连接

让我们用一个简单的例子来

理解这一点

我们手头有一张图像,

我们的神经网络的任务

是识别该图像

是猫还是码头

的图像,我们解决的方法 通过获取

该图像的像素

并将其馈送到最低层(也

称为输入层),

该层处理此信息

,然后将其传递到下一个较

高层,

此信息流

通过多个层重复,

直到到达顶部 该层有

两个神经元,

一个对应于一只狗,另一个

对应于一只猫,

这个神经元被激活得更多

在顶层

,本质上

是这个神经网络预测的最终答案,

所以这就是我们现在如何使用这个神经

网络来获得

答案,人们可能想

知道我们如何教或训练这个神经

网络

来教它我们喂一堆 在

已知示例中,当网络给出错误答案时,我们已经

通过该神经网络知道

答案,我们利用这个机会

通过网络反向传播信息

反向传播本质上是一种

反馈机制

,我们使用它来纠正层之间的边缘权重

例如 下一次它更有

可能给我们一个正确的答案,

这就像从你的

错误

中学习另一方面,当神经

网络给出一个正确的答案时,

我们奖励它,我们

通过

与 我们奖励

孩子做正确的事

反向传播技术

从你自己的错误中学习

是什么让神经网络 它是如此

强大,几乎可以学习

任何东西,

所以现在我们

对人工智能和神经网络的基本概念有了一个概念,让我们

谈谈它将如何帮助我们的社会变得

更幸福

空气已经在我们周围的应用中使用了

人工智能正在帮助我们 当您在线观看视频时,

通过自动驾驶汽车和智能

交通信号灯

让道路更安全 人工智能通常会

在您发送短信时根据您的偏好帮助您决定观看哪个视频 人工智能甚至可以

通过这些应用程序帮助您起草该短信

通过自动化平凡的任务让我们更有效率,

通过

腾出时间做我们最喜欢的事情来让我们更快乐

[音乐]

但我们在 21 世纪继续面临的一些最具挑战性的

问题是

饥饿气候变化或 即使是文盲

,因为我们的箭袋里有人工智能,

我们应该期待甚至理所当然地

要求它可以帮助我们解决这些

难题

让我们先来谈谈 hun 德国大约有 6.9

亿人

仍然每晚空腹睡觉

我们处于 21 世纪 我们声称

自己是一个现代文明

,所以如此

多的人因缺乏食物而受苦是不可接受的,这

是导致饥饿的主要原因

不仅是洪水和地震等自然灾害

,还

包括战争等人类造成

的灾难 当灾难袭来时,大量

人口无法获得

食物和住所等基本必需品

救援往往比

救援物资的供应问题更大,

因此我们与联合国合作

开发了人工智能技术,可以

通过分析受灾地区的航拍图像来帮助向受灾地区部署援助,

我们分析航拍图像以

确定受灾地区的

位置 受损害影响最大,

因此急需援助,

我们解决了这个问题 问题 通过使用深度

学习

来比较

一个地区的灾难前后的图像

当我们比较这两个图像时,我们

能够确定由于灾难而造成最大破坏的地方

我们使用两步过程来做到这一点

第一步

是在确定建筑物后检测

给定

区域中的所有建筑物

第二步

是确定这些建筑物

中的哪些被灾难损坏

这是左侧海地地震的示例,

我们可以看到建筑物

被识别为损坏

和 在右侧,我们可以看到

建筑物被确定为未受损

我们可以看到这项技术能够

做出有能力

甚至比人类做出的预测更好的预测

当这些信息汇总

在一起时,会

突出显示受灾难伤害的区域

这些数据 这项技术

对于向

急需这种援助(

如食品和药品)

的人们提供援助非常有用 正在帮助

解决

饥饿问题,但人工智能的潜力

远远超出医疗保健和医疗

保健

教育等领域

我们已经在糖尿病视网膜病变领域看到这种情况发生的每一个人

,不幸的是,这一直是导致

失明的主要原因

他们的健康状况因

使用人工智能而下降 我们现在拥有的诊断

技术

的性能至少与

医生一样

好 这使得诊断更具

可扩展性和可访问性

此应用程序是

人类人工智能合作

的完美示例,人工智能可以完成繁重的工作并

解放我们 医生

花更多时间与患者互动和

帮助他们的患者,

我设想人工智能 沿着类似

的人类人工智能伙伴关系工作

,人类和人工智能一起工作

教育教育是每个社会的基石

毕竟这有助于我们的

孩子成长为有能力的公民,

不幸的是,仍有数百万

儿童

无法接受正规教育

如果 ei 可以

通过

根据每个孩子的优势和需求为每个孩子提供个性化的学习体验,从而使教育更容易获得,

每个孩子都是独一无二的,应该

获得为他们量身定制的高质量教育,无论是

农业还是教育 或者

医疗保健可以从人工智能中受益匪浅

想象智能手机可以帮助

农民计划和种植农作物 获取收成

信息

预报天气预报 降雨量,

甚至获得农业供应

如果技术

不能真正提升被压抑的人并

丰富他们的生活,那它还有什么用

呢? 人工智能不是科幻小说,

它是一个强大的工具 已经在

这里让我们的生活更轻松、更快乐

,我要说的是,

我们不能让非理性的恐惧阻止

我们使用人工智能来造福人类

,就像 18 世纪发明的蒸汽机

或 19 世纪的电话发明者一样,

人工智能拥有 显着

改善我们生活的潜力

EI 将帮助数百万人

摆脱贫困,

但就像蒸汽机和

电话

一样,它只会在我们创造它的时候发挥作用,

因此我们每个人都需要确定

这样的生产力 机会

和利用人工智能来帮助治愈世界

让我们一起努力,用人工智能让

我们的世界

更快乐,更和平,非常感谢

[鼓掌]

非常

感谢