How AI Could Empower Any Business | Andrew Ng | TED

譯者: Val Zhang 審譯者: Shelley Tsang 曾雯海

When I think about the rise of AI,

人工智慧的興起總讓我 聯想起識字率的普及。

I'm reminded by the rise of literacy.

A few hundred years ago,

數百年前，

many people in society thought

社會的多數認為 識字並非必要技能。

that maybe not everyone needed to be able to read and write.

Back then, many people were tending fields or herding sheep,

那時很多人從事農業、畜牧，

so maybe there was less need for written communication.

也許不太需要書面溝通。

And all that was needed

只要祭司和僧侶們能夠閱讀聖書，

was for the high priests and priestesses and monks

to be able to read the Holy Book,

and the rest of us could just go to the temple or church

我們其他人只需前往宗教的殿堂，

or the holy building

and sit and listen to the high priest and priestesses read to us.

坐著聽祭司們為我們而讀；

Fortunately, it was since figured out that we can build a much richer society

幸好後來意識到識字率的提升 能共創更富裕的社會。

if lots of people can read and write.

Today, AI is in the hands of the high priests and priestesses.

如今，人工智慧掌握 在崇高的祭司們手中──

These are the highly skilled AI engineers,

他們是技術高超的工程師，

many of whom work in the big tech companies.

多為大型科技公司效力，

And most people have access only to the AI that they build for them.

多數人能接觸的人工智慧 均由上述的少數人建構。

I think that we can build a much richer society

我認為我們能建立更富裕的社會──

if we can enable everyone to help to write the future.

若我們賦能每個人共同編寫未來。

But why is AI largely concentrated in the big tech companies?

為何人工智慧集中於 大型科技公司？

Because many of these AI projects have been expensive to build.

因為多數人工智慧專案 的建造成本很高。

They may require dozens of highly skilled engineers,

可能要數十位技術高超的工程師，

and they may cost millions or tens of millions of dollars

可能會花費數百萬、千萬美元

to build an AI system.

以建構一套人工智慧系統。

And the large tech companies,

特別是擁有幾十億計用戶 的大型科技公司，

particularly the ones with hundreds of millions

or even billions of users,

have been better than anyone else at making these investments pay off

在人工智慧的投資報酬率上傲視群雄，

because, for them, a one-size-fits-all AI system,

因為一套通用的人工智慧系統，

such as one that improves web search

像是：優化搜尋結果

or that recommends better products for online shopping,

或網購的商品推薦，

can be applied to [these] very large numbers of users

適用於大量用戶以產生鉅額營收。

to generate a massive amount of revenue.

But this recipe for AI does not work

一旦離開科技或網路業， 人工智慧的規模經濟便難以實現，

once you go outside the tech and internet sectors to other places

在其他產業，多數情況下，

where, for the most part,

there are hardly any projects that apply to 100 million people

幾乎沒有任何專案適用於一億人，

or that generate comparable economics.

或產生相對應的經濟價值。

Let me illustrate an example.

讓我舉個例子：

Many weekends, I drive a few minutes from my house to a local pizza store

週末我常開車到 離家不遠的本地比薩店，

to buy a slice of Hawaiian pizza

向比薩店的老闆買片夏威夷披薩。

from the gentleman that owns this pizza store.

And his pizza is great,

他的披薩很棒，

but he always has a lot of cold pizzas sitting around,

但總有些比薩無人問津， 而週末有些則供不應求。

and every weekend some different flavor of pizza is out of stock.

But when I watch him operate his store,

但看著他的營運，我感到興奮，

I get excited,

because by selling pizza,

因為賣披薩的同時， 也在產生數據。

he is generating data.

And this is data that he can take advantage of

若能運用人工智慧 就能利用數據創造價值。

if he had access to AI.

AI systems are good at spotting patterns when given access to the right data,

若獲得合適的數據， 人工智慧便擅於辨識出模式，

and perhaps an AI system could spot if Mediterranean pizzas sell really well

若人工智慧發現： 週五晚上地中海披薩銷量高，

on a Friday night,

maybe it could suggest to him to make more of it on a Friday afternoon.

也許它能建議在週五下午備好足量。

Now you might say to me, "Hey, Andrew, this is a small pizza store.

你可能會跟我說： 「安德魯，這是家小披薩店，

What's the big deal?"

有什麼大不了的？」

And I say, to the gentleman that owns this pizza store,

但我認為，對比薩店的老闆而言，

something that could help him improve his revenues

若能提高每年收入幾千美元，

by a few thousand dollars a year, that will be a huge deal to him.

這對他來說意義重大。

I know that there is a lot of hype about AI's need for massive data sets,

我知道許多炒作總說── 人工智慧需要巨量數據，

and having more data does help.

數據的確多多益善。

But contrary to the hype,

但與炒作相反，

AI can often work just fine

即便少量的數據， 人工智慧也能發揮價值，

even on modest amounts of data,

such as the data generated by a single pizza store.

就像：一家比薩店所產生的數據。

So the real problem is not

真正的問題不是 比薩店沒有足夠的數據；

that there isn’t enough data from the pizza store.

The real problem is that the small pizza store

真正的問題是──

小比薩店的營收再好，也不足以支撐

could never serve enough customers

to justify the cost of hiring an AI team.

雇用人工智慧團隊的投資報酬率。

I know that in the United States

據我所知：美國約 五十萬間獨立餐廳，

there are about half a million independent restaurants.

And collectively, these restaurants do serve tens of millions of customers.

這些餐廳服務合計 約數千萬名客戶。

But every restaurant is different with a different menu,

但是每家餐廳都不同──

從菜單、客戶到記錄銷售的方式，

different customers, different ways of recording sales

that no one-size-fits-all AI would work for all of them.

沒有一套人工智慧 適用於所有餐廳。

What would it be like if we could enable small businesses

若賦能小型、尤其是本地企業， 運用人工智慧，將會如何？

and especially local businesses to use AI?

Let's take a look at what it might look like

讓我們以 T 恤的製造零售商為例，

at a company that makes and sells T-shirts.

I would love if an accountant working for the T-shirt company

我很期待 T 恤公司的會計師 能用人工智慧預測需求，

can use AI for demand forecasting.

Say, figure out what funny memes to prints on T-shirts

像：為帶動銷售，找出在 T 恤上 該印什麼有趣的迷因，

that would drive sales,

by looking at what's trending on social media.

藉由觀察社群媒體的趨勢。

Or for product placement,

或商品陳列──

why can’t a front-of-store manager take pictures of what the store looks like

何不由店經理拍下 商店的樣貌給人工智慧看，

and show it to an AI

and have an AI recommend where to place products to improve sales?

並由其推薦商品陳列 位置，以提高銷量？

Supply chain.

或用在供應鏈──

Can an AI recommend to a buyer whether or not they should pay 20 dollars

人工智慧能建議採購：

現在是否應該 支付一碼布二十美元，

per yard for a piece of fabric now,

or if they should keep looking

或者應該繼續詢價

because they might be able to find it cheaper elsewhere?

因為也許能找到更便宜的貨源？

Or quality control.

或運用於品質管理──

A quality inspector should be able to use AI

品檢員應能運用人工智慧

to automatically scan pictures of the fabric they use to make T-shirts

自動掃描 T 恤布料纖維的照片，

to check if there are any tears or discolorations in the cloth.

檢查布料是否有裂縫或變色。

Today, large tech companies routinely use AI to solve problems like these

大型科技公司已習於運用 人工智慧解決類似的問題，

and to great effect.

且效果顯著。

But a typical T-shirt company or a typical auto mechanic

但是典型的 T 恤公司、修車行、

or retailer or school or local farm

零售商、學校或本地農場，

will be using AI for exactly zero of these applications today.

至今尚未能運用人工智慧。

Every T-shirt maker is sufficiently different from every other T-shirt maker

每家 T 恤製造商都有些許不同，

that there is no one-size-fits-all AI that will work for all of them.

沒有一套通用的人工智慧 適用於所有廠商。

And in fact, once you go outside the internet and tech sectors

實際上，一旦踏出 網路和科技領域，

in other industries, even large companies

在其他行業，即便是大公司，

such as the pharmaceutical companies,

比如：製藥公司、汽車製造商、醫院，

the car makers, the hospitals,

also struggle with this.

也還在努力中。

This is the long-tail problem of AI.

這是人工智慧（客製）應用的長尾現象。

If you were to take all current and potential AI projects

若將既有及潛在的人工智慧專案

and sort them in decreasing order of value and plot them,

並依價值遞減排序繪製，

you get a graph that looks like this.

會得到這樣的圖表。

Maybe the single most valuable AI system

也許規模經濟價值最高 的是網路廣告推薦。

is something that decides what ads to show people on the internet.

Maybe the second most valuable is a web search engine,

其次是網路搜尋引擎，

maybe the third most valuable is an online shopping product recommendation system.

再者是網購商品推薦。

But when you go to the right of this curve,

但沿著這條曲線向右，

you then get projects like T-shirt product placement

才是 T 恤的商品陳列、需求預測，

or T-shirt demand forecasting or pizzeria demand forecasting.

或比薩店需求預測。

And each of these is a unique project that needs to be custom-built.

這些都是獨特的專案，需要客製。

Even T-shirt demand forecasting,

即使是 T 恤需求預測，

if it depends on trending memes on social media,

若其取決於：社群媒體的話題夯物，

is a very different project than pizzeria demand forecasting,

便與比薩店的需求預測 有本質性的差異，

if that depends on the pizzeria sales data.

若後者取決於：比薩店的銷售數據。

So today there are millions of projects

現今有數百萬個專案 坐落於分佈的尾端，

sitting on the tail of this distribution that no one is working on,

無人問津，但其 合計價值不可小覷。

but whose aggregate value is massive.

So how can we enable small businesses and individuals

如何賦能小型企業和個人去建構 對其重要的人工智慧系統？

to build AI systems that matter to them?

For most of the last few decades,

在過去幾十年裡， 若想建立人工智慧系統，

if you wanted to build an AI system, this is what you have to do.

別無他法──你得寫數頁的程式。

You have to write pages and pages of code.

And while I would love for everyone to learn to code,

雖然我樂見人人學寫程式，

and in fact, online education and also offline education

而在實體和線上課程 學寫程式的人數已破歷史新高，

are helping more people than ever learn to code,

unfortunately, not everyone has the time to do this.

可惜不是每個人都有時間學程式。

But there is an emerging new way

但有種新興的方式

to build AI systems that will let more people participate.

建構能讓更多人參與 的人工智慧系統。

Just as pen and paper,

相對於石碑、鑿子， 紙、筆是大幅領先的技術──

which are a vastly superior technology to stone tablet and chisel,

were instrumental to widespread literacy,

是識字率得以普及的要素，

there are emerging new AI development platforms

新興的人工智慧開發平台

that shift the focus from asking you to write lots of code

將焦點從「寫程式」轉為「提供數據」。

to asking you to focus on providing data.

And this turns out to be much easier for a lot of people to do.

事實證明，這對許多人來說更容易。

Today, there are multiple companies working on platforms like these.

如今有許多公司 正開發類似的平台，

Let me illustrate a few of the concepts using one that my team has been building.

以我的團隊運用的概念 向各位舉個例：

Take the example of an inspector

品檢員希望人工智慧 協助檢測織物中的瑕疵，

wanting AI to help detect defects in fabric.

An inspector can take pictures of the fabric

品檢員能拍下布料的照片

and upload it to a platform like this,

並將其上傳到這樣的平台，

and they can go in to show the AI what tears in the fabric look like

他們能透過標註方框 告訴人工智慧「裂縫」的樣子。

by drawing rectangles.

And they can also go in to show the AI

也能透過標註方框 告訴人工智慧「變色」的樣子。

what discoloration on the fabric looks like

by drawing rectangles.

So these pictures,

這些由品檢員標註的 綠色、粉紅色方框的照片，

together with the green and pink rectangles

that the inspector's drawn,

are data created by the inspector

就是品檢員建立的數據，

to explain to AI how to find tears and discoloration.

告訴人工智慧如何找出裂縫和變色。

After the AI examines this data,

人工智慧檢查這些數據後，

we may find that it has seen enough pictures of tears,

可能發現「裂縫」的照片已足量，

but not yet enough pictures of discolorations.

但還沒有足夠「變色」的照片。

This is akin to if a junior inspector had learned to reliably spot tears,

這類似初級品檢員 已學會準確地揪出「裂縫」，

but still needs to further hone their judgment about discolorations.

但仍需進一步磨練 對「變色」的判斷力。

So the inspector can go back and take more pictures of discolorations

因此品檢員能再為人工智慧 拍更多「變色」的照片，

to show to the AI,

to help it deepen this understanding.

幫助其更加理解。

By adjusting the data you give to the AI,

透過調整提供給人工智慧的數據，

you can help the AI get smarter.

你能幫助人工智慧變得更聰明。

So an inspector using an accessible platform like this

品檢員運用 像這樣操作友善的平台，

can, in a few hours to a few days,

能在幾小時到幾天內，

and with purchasing a suitable camera set up,

並搭配合適的相機裝置，

be able to build a custom AI system to detect defects,

能建立客製的 人工智慧系統以檢測瑕疵，

tears and discolorations in all the fabric

整個 T 恤工廠的所有 織物的裂縫和變色均適用。

being used to make T-shirts throughout the factory.

And once again, you may say,

同樣的，你可以說：

"Hey, Andrew, this is one factory.

「安德魯，這只是 一家工廠，有何大不了？」

Why is this a big deal?"

And I say to you,

我會說：

這對工作品質被提升的 品檢員而言意義非凡；

this is a big deal to that inspector whose life this makes easier

and equally, this type of technology can empower a baker to use AI

同樣地，這類的技術 能讓烘焙師運用人工智慧

to check for the quality of the cakes they're making,

檢查蛋糕的品質，

or an organic farmer to check the quality of the vegetables,

或讓有機農民檢驗蔬菜的品質，

or a furniture maker to check the quality of the wood they're using.

或讓家具製造商檢查木材的品質。

Platforms like these will probably still need a few more years

這類的平台可能還需 數年的開發和優化，

before they're easy enough to use for every pizzeria owner.

才能讓每位比薩店 經營者都容易上手操作，

But many of these platforms are coming along,

但這些平台都在持續精進，

and some of them are getting to be quite useful

甚至對熟悉科技者，只需一點訓練． 部分平台已能堪稱實用。

to someone that is tech savvy today,

with just a bit of training.

But what this means is that,

這也意味著──

rather than relying on the high priests and priestesses

不再得依靠崇高的祭司們 為眾人編寫人工智慧系統，

to write AI systems for everyone else,

we can start to empower every accountant,

我們能開始賦能給每位 會計師、店長、採購和品檢員，

every store manager,

every buyer and every quality inspector to build their own AI systems.

建立對應的人工智慧系統。

I hope that the pizzeria owner

我希望比薩店老闆 和許多這類的小型企業主，

and many other small business owners like him

will also take advantage of this technology

也能善用這項技術，

because AI is creating tremendous wealth

因為人工智慧正在創造鉅額財富，

and will continue to create tremendous wealth.

並將繼續創造鉅額財富，

And it's only by democratizing access to AI

唯有透過將人工智慧賦能於民

that we can ensure that this wealth is spread far and wide across society.

才能確保社會上財富流動的廣度。

Hundreds of years ago.

數百年前，我想沒人能體認──

I think hardly anyone understood the impact

that widespread literacy will have.

識字率普及的影響。

Today, I think hardly anyone understands

如今，我想沒人能體認──

the impact that democratizing access to AI will have.

普及運用人工智慧的影響。

Building AI systems has been out of reach for most people,

建構人工智慧系統對 多數人而言遙不可及，

but that does not have to be the case.

但我們能夠改變現狀。

In the coming era for AI,

人工智慧時代即將到來，

we’ll empower everyone to build AI systems for themselves,

我們將賦能每個人 建立對應的人工智慧，

and I think that will be incredibly exciting future.

我想那會是令人興奮的未來。

Thank you very much.

謝謝大家。

(Applause)

（掌聲）