Transcript

Transcript of YouTube Video: How AI Could Empower Any Business | Andrew Ng | TED

Transcript of YouTube Video: How AI Could Empower Any Business | Andrew Ng | TED

The following is a summary and article by AI based on a transcript of the video "How AI Could Empower Any Business | Andrew Ng | TED". Due to the limitations of AI, please be careful to distinguish the correctness of the content.

Article By AIVideo Transcript
00:04

When I think about the rise of AI,

00:07

I'm reminded by the rise of literacy.

00:10

A few hundred years ago,

00:11

many people in society thought

00:13

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

00:17

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

00:20

so maybe there was less need for written communication.

00:23

And all that was needed

00:24

was for the high priests and priestesses and monks

00:26

to be able to read the Holy Book,

00:28

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

00:31

or the holy building

00:32

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

00:35

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

00:39

if lots of people can read and write.

00:42

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

00:46

These are the highly skilled AI engineers,

00:48

many of whom work in the big tech companies.

00:51

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

00:55

I think that we can build a much richer society

00:58

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

01:03

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

01:08

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

01:11

They may require dozens of highly skilled engineers,

01:14

and they may cost millions or tens of millions of dollars

01:17

to build an AI system.

01:19

And the large tech companies,

01:20

particularly the ones with hundreds of millions

01:22

or even billions of users,

01:24

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

01:28

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

01:33

such as one that improves web search

01:35

or that recommends better products for online shopping,

01:38

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

01:41

to generate a massive amount of revenue.

01:44

But this recipe for AI does not work

01:47

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

01:52

where, for the most part,

01:53

there are hardly any projects that apply to 100 million people

01:57

or that generate comparable economics.

02:00

Let me illustrate an example.

02:03

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

02:09

to buy a slice of Hawaiian pizza

02:11

from the gentleman that owns this pizza store.

02:14

And his pizza is great,

02:15

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

02:19

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

02:23

But when I watch him operate his store,

02:25

I get excited,

02:27

because by selling pizza,

02:29

he is generating data.

02:31

And this is data that he can take advantage of

02:34

if he had access to AI.

02:37

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

02:43

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

02:47

on a Friday night,

02:48

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

02:53

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

02:56

What's the big deal?"

02:58

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

03:01

something that could help him improve his revenues

03:03

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

03:08

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

03:14

and having more data does help.

03:17

But contrary to the hype,

03:19

AI can often work just fine

03:21

even on modest amounts of data,

03:23

such as the data generated by a single pizza store.

03:26

So the real problem is not

03:28

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

03:30

The real problem is that the small pizza store

03:33

could never serve enough customers

03:34

to justify the cost of hiring an AI team.

03:39

I know that in the United States

03:41

there are about half a million independent restaurants.

03:44

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

03:48

But every restaurant is different with a different menu,

03:51

different customers, different ways of recording sales

03:53

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

03:58

What would it be like if we could enable small businesses

04:01

and especially local businesses to use AI?

04:05

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

04:07

at a company that makes and sells T-shirts.

04:10

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

04:14

can use AI for demand forecasting.

04:16

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

04:19

that would drive sales,

04:20

by looking at what's trending on social media.

04:23

Or for product placement,

04:25

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

04:29

and show it to an AI

04:30

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

04:34

Supply chain.

04:35

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

04:39

per yard for a piece of fabric now,

04:41

or if they should keep looking

04:43

because they might be able to find it cheaper elsewhere?

04:46

Or quality control.

04:47

A quality inspector should be able to use AI

04:50

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

04:55

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

04:59

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

05:04

and to great effect.

05:06

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

05:11

or retailer or school or local farm

05:15

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

05:19

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

05:24

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

05:28

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

05:33

in other industries, even large companies

05:35

such as the pharmaceutical companies,

05:37

the car makers, the hospitals,

05:39

also struggle with this.

05:42

This is the long-tail problem of AI.

05:46

If you were to take all current and potential AI projects

05:50

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

05:55

you get a graph that looks like this.

05:57

Maybe the single most valuable AI system

05:59

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

06:02

Maybe the second most valuable is a web search engine,

06:05

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

06:09

But when you go to the right of this curve,

06:12

you then get projects like T-shirt product placement

06:15

or T-shirt demand forecasting or pizzeria demand forecasting.

06:20

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

06:24

Even T-shirt demand forecasting,

06:26

if it depends on trending memes on social media,

06:29

is a very different project than pizzeria demand forecasting,

06:34

if that depends on the pizzeria sales data.

06:37

So today there are millions of projects

06:39

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

06:43

but whose aggregate value is massive.

06:46

So how can we enable small businesses and individuals

06:49

to build AI systems that matter to them?

06:52

For most of the last few decades,

06:54

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

06:58

You have to write pages and pages of code.

07:00

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

07:03

and in fact, online education and also offline education

07:06

are helping more people than ever learn to code,

07:09

unfortunately, not everyone has the time to do this.

07:13

But there is an emerging new way

07:16

to build AI systems that will let more people participate.

07:20

Just as pen and paper,

07:22

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

07:26

were instrumental to widespread literacy,

07:29

there are emerging new AI development platforms

07:32

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

07:35

to asking you to focus on providing data.

07:39

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

07:43

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

07:47

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

07:51

Take the example of an inspector

07:54

wanting AI to help detect defects in fabric.

07:58

An inspector can take pictures of the fabric

08:00

and upload it to a platform like this,

08:03

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

08:07

by drawing rectangles.

08:09

And they can also go in to show the AI

08:11

what discoloration on the fabric looks like

08:14

by drawing rectangles.

08:16

So these pictures,

08:17

together with the green and pink rectangles

08:19

that the inspector's drawn,

08:21

are data created by the inspector

08:23

to explain to AI how to find tears and discoloration.

08:28

After the AI examines this data,

08:30

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

08:32

but not yet enough pictures of discolorations.

08:35

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

08:39

but still needs to further hone their judgment about discolorations.

08:43

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

08:47

to show to the AI,

08:48

to help it deepen this understanding.

08:50

By adjusting the data you give to the AI,

08:53

you can help the AI get smarter.

08:56

So an inspector using an accessible platform like this

09:00

can, in a few hours to a few days,

09:03

and with purchasing a suitable camera set up,

09:07

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

09:11

tears and discolorations in all the fabric

09:13

being used to make T-shirts throughout the factory.

09:16

And once again, you may say,

09:19

"Hey, Andrew, this is one factory.

09:22

Why is this a big deal?"

09:23

And I say to you,

09:25

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

09:28

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

09:32

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

09:35

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

09:39

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

09:44

Platforms like these will probably still need a few more years

09:47

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

09:51

But many of these platforms are coming along,

09:53

and some of them are getting to be quite useful

09:56

to someone that is tech savvy today,

09:58

with just a bit of training.

10:00

But what this means is that,

10:02

rather than relying on the high priests and priestesses

10:04

to write AI systems for everyone else,

10:07

we can start to empower every accountant,

10:10

every store manager,

10:11

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

10:17

I hope that the pizzeria owner

10:19

and many other small business owners like him

10:22

will also take advantage of this technology

10:24

because AI is creating tremendous wealth

10:28

and will continue to create tremendous wealth.

10:30

And it's only by democratizing access to AI

10:33

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

10:39

Hundreds of years ago.

10:41

I think hardly anyone understood the impact

10:44

that widespread literacy will have.

10:47

Today, I think hardly anyone understands

10:50

the impact that democratizing access to AI will have.

10:54

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

10:58

but that does not have to be the case.

11:01

In the coming era for AI,

11:03

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

11:06

and I think that will be incredibly exciting future.

11:10

Thank you very much.

11:11

(Applause)