Magic and Wonder in the Age of AI | David Kwong | TED

I am both a magician

and a "New York Times" crossword puzzle constructor,

and I think that all magic tricks are puzzles.

Puzzles designed to fool your brain.

I'm also a self-described word nerd,

and I was thinking the other day about the word "wonder,"

and the meaning of the word wonder.

What is wonder?

Well, I'm not going to bore you with etymology for two reasons.

Number one, because being bored by etymology is impossible,

because etymology is amazing.

(Laughter)

And number two, I looked it up.

And according to the online etymological dictionary,

the word wonder is ultimately of unknown origin,

which is pretty perfect, right?

It's just one more thing for us to wonder about.

Now this brings me to an interesting point.

Today, when we use wonder as a verb, it can be pretty humdrum.

Like, "I wonder what she had for lunch."

Or as an adjective, same thing.

"Mm, this cake is wonderful."

That's just, like, pretty good cake.

It's only when we use wonder as a noun that we get a sense of its true meaning.

"The Great Pyramid of Giza,

the last remaining of the seven ancient wonders of the world."

A wonder is an experience or an object that defies explanation,

that blows our mind,

that stirs within us that deep and undeniable sense of wonder.

A wonder, therefore, is something that when we see it,

it makes us question what is possible.

We see something that is impossible, and there it is right in front of us.

It's that disconnect,

that yawning gap between shouldn't be and is

that inspires wonder.

And of course, when we think of wonder and wondrous times,

it often helps to cast our minds back

to an earlier age when, we assume, man's capacity for wonder was greater.

I mentioned the Great Pyramid of Giza before, and there was a time when,

for an ancient traveler winding their way up the Nile

and first catching sight of a perfectly symmetrical artificial mountain

rising hundreds of feet above the desert floor,

that it would have been simply unimaginable.

It shouldn't be, but it is.

And now, you know, we have the Burj Khalifa.

It's way bigger.

And though most of us who aren't architects and engineers

wouldn't say we know exactly how they built the Burj,

we're pretty confident that somebody does with architecty-engineery stuff.

It's possible.

So that leads me to today's question.

Has too much technology robbed us of wonder?

Have we traded in magic and mystery for cold, hard logic and logistics?

Can we today, with our scientific sophistication,

ever recapture our ancestors' capacity for wonder?

And I'm glad you asked, rhetorically.

Because I would answer with a resounding yes.

I would say that not only can we experience wonder

despite our modern marvels and technological advances,

but that we can experience more wonder because of them.

We live in a world more wondrous

than our ancestors ever could have imagined,

and technology isn't the barrier to unlocking wonder.

It is the key.

OK, look at this.

Rainbows.

We all love rainbows.

How did those pretty colors get up in the sky?

Now look at this.

The Pillars of Creation.

As photographed by the Hubble Telescope in 1995.

The Pillars of Creation are 7,000 light years from Earth

and five light years tall.

Five light years tall.

Take that, rainbows.

(Laughter)

And no Hubble telescope, and no Pillars of Creation.

I mean, they'd still be there, of course, but we would never be able to see them.

Advancements in technology increase our opportunities for wonder.

As a magician, I have the coolest job in the world.

I get to stand in a long line of men and women,

of priests and shamans,

of wizards and witches

who stand on the precipice of technological advancement

and look out from the cutting edge of mankind's achievement

into the eternal expanse of the unknown

and say, "I can totally use this to screw with people's minds."

Which brings us to AI.

You're all familiar with ChatGPT.

Have you played with it?

Yes.

Well, I've been experimenting with ChatGPT,

and I've discovered that there's a fun little party trick that it can do

if you give it enough information.

It can make certain deductions.

And I want to show you what I'm talking about.

I have ChatGPT right here.

And let's warm it up with, how about this,

please write a haiku about a TED talk in Paris.

Here we go.

"Paris TED talk airs

Ideas dance with Eiffel's grace

Wisdom blooms like spring."

That's nice.

(Laughter)

Alright.

I want to try something with someone from the audience.

Do I have a volunteer who might want -- it's going to be painless, I promise.

Are you willing to join me?

What is your name?

François.

That's a unique name here in Paris.

(Laughter)

Please join me, let's give François ...

(Applause)

Thank you.

Hello, hello.

François, I'm going to put some information about you

into ChatGPT, if that's OK.

Alright, let's have some fun with it.

So I am here with Francois.

F-r-a-n-c-o-i-s, right?

I just wanted to make sure I got that right.

Where are you from originally?

Yes, there's a little thing,

but I don't know how to type the little thing.

(Laughter)

Where are you from originally?

Right here, from Paris.

What is your favorite food?

Francois: Very French, it's confit.

David Kwong: Confit? Yes.

C-o-n-f-i-t, yes, I do know that.

It's very fancy.

Do you have any pets?

Francois: No.

DK: Did you grow up with any pets?

Francois: Yes.

DK: What was your pet?

Francois: A cat.

DK: A cat named?

Francois: Pistache.

(Laughter)

DK: You're going to have to spell that.

Francois: Pistachio.

DK: Pistachio!

With an E? Pistache?

Francois: Yes.

DK: Pistache.

Is there a little thing? There's no little thing, OK.

And are you right-handed or left-handed?

Francois: Right.

DK: You are right-handed.

He is right-handed.

OK, so we have some information, watch this.

Please write a sea shanty about Francois.

(Laughter)

Verse one, "Now gather round, me hearties bold,

I'll tell a tale so grand and old

Of Francois from Paris town

Whose adventures will astound."

That's great. What else?

"Oh, Francois,

A man so fine with a cat named Pistache

Were they intertwined?

He sailed the seas with the spirit high, a true sailor reaching for the sky."

That's very good.

Alright.

I would like to play a little game with you.

Alright?

Thank you so much for volunteering.

I have here an envelope -- this is still going.

This is a long sea shanty.

(Laughter)

"Raise your voices, sing his praise --"

OK, this is my prediction envelope.

We'll come back to this in just a little bit.

I'm going to put this right here, OK.

Francois, we're going to play a game of which hand.

I have a coin here, OK?

In a moment you're going to put it behind your back,

and you're going to put it in one of your two fists OK?

But let me add another layer here.

Can you please think of your favorite color?

Anything.

What do you have?

Tell me please. Blue.

And then think of another color.

Anything, really.

Go ahead, tell me.

OK, blue and red that's perfect, great.

Can you please come forward with your two hands,

put the coin in one of them.

If the coin is in your left hand, I want you to think of blue.

And if it's in your right hand, I want you to think of red.

And let that color just rise up throughout you

and project it into the room as I type into ChatGPT.

Francois has hidden a coin in one of two hands.

I've assigned the left hand the color blue

and the right hand the color red.

Which hand has the coin?

Let's see what ChatGPT says.

"As an AI language model,

I can't exactly determine what happens in the physical world,

but based on the information provided,

Francois would likely hide the coin in the blue hand."

Are you thinking of blue?

Yes, indeed, show everybody the blue hand has the coin.

Very nice, OK.

(Applause)

Now, of course, that could just be 50-50, right?

So let's try it again.

Would you put it behind your back again?

And could you please name for me any animal you like?

Francois: A cat.

DK: A cat, good.

And then maybe something completely different,

Like from the zoo.

A lion, wonderful.

Would you please come out with your two fists again?

Go ahead, mix it up, make it difficult.

Don't put it in your back pocket.

Yes, OK, OK.

So just like before, I want you to think, alright?

If the coin is in your right hand, I want you to think of cat.

If it's in your left hand, I want you to think of lion, alright?

I want the spirit of this animal to take over.

Now don't jump up and down, OK?

But let the spirit project that animal into the room.

And here we are.

Francois is hiding a coin

in one of his two hands.

I've assigned the right hand the cat

and the left hand the lion.

Which hand has the coin?

And ChatGPT says:

"Based on the information provided,

Francois would hide the coin in the lion hand."

Are you thinking of lion?

Would you show everybody the lion hand?

Look at that.

Two for two, that's good.

That's good.

Let's try one more time.

How about fruit and vegetables?

Go ahead and name two.

Two things, anything you like.

Francois: Apple and strawberry.

DK: What was the first one?

Francois: Apple.

DK: Apple and strawberry.

My French is not very good.

(Laughter)

Apple and strawberry.

Wonderful.

OK, come out like this.

Alright, so if the coin is in your right hand,

I want you to think of apple.

If it's in your left hand, I want you to think of strawberry.

And I want you to think about what that fruit tastes like.

And really project that feeling into the room as I type here.

Francois has a coin in one of his hands.

I've assigned apple to the right hand.

And strawberry to the left.

Which hand has the coin?

And ChatGPT says:

"Based on the previous information,

Francois would likely hide the coin in the apple hand."

Were you thinking of apple?

You don't look very happy about this.

But show everybody that you were thinking of apple.

And it's in the apple hand.

It is indeed right there, very nice, good.

(Applause)

Now ChatGPT just had all the fun.

So now it's my turn to try to have a little deduction here, OK?

So I would like you please to, in a moment, think of a country.

Don't say it out loud.

There are nearly 200 countries in the world, think of one.

You got one?

Look right here.

Look right here, OK.

And let's see, so it was apple, alright, the lion.

Mhm, OK.

I have an idea.

I have an idea.

Yes or no.

Is it Cuba?

No it's not, but that's OK, good.

Alright, interesting.

I did not get the country initially,

but I am going to ask ChatGPT

a few more questions to see if it can make some deductions here

about what you're thinking.

So let's do this.

Name a musician from 1899 --

Born, born --

Name a musician born in 1899.

How about this?

A world leader, born in AD --

I'm going to go way back -- AD 37,

an athlete from 1942,

and a writer,

let's do a British writer from 1865.

Here we are.

"Certainly, here are the individuals.

Musician born in 1899, Duke Ellington."

Very good.

"A Roman emperor, born in AD 37, Nero.

An athlete born in 1942, Muhammad Ali, born Cassius Clay,

and for the British writer," good,

"born in 1865, Rudyard Kipling."

Do these people from history mean anything to you?

No, but that's OK.

That's OK, they will in just a minute.

Because I had a prediction before in this envelope.

And I'm going to show you now what it says.

It says,

"I will not get the country initially."

(Laughter)

Thank you very much.

This initially is a double entendre.

How's my French, entendre?

Terrible, right?

Entendre.

It means two things, initially.

Let's go back to ChatGPT.

Please list the initials of the people

I just asked about.

Are you ready?

What was the country you were thinking of before?

Francois: Denmark.

DK: Denmark, watch this.

ChatGPT says,

"Duke Ellington, DE,

Nero, N

Muhammad Ali, MA and Rudyard Kipling, RK."

That's D-E-N-M-A-R-K, that's Denmark.

Let's give Francois a big round of applause.

Thank you so much.

(Applause)

That was ...

(Applause)

That was fun and magical.

Thank you so much.

And that leads us back to wonder.

We said at the beginning that wonder was questioning what is possible.

And here's the thing.

It's the questioning that's important.

The Pillars of Creation,

those unthinkably vast clouds of stardust.

They existed long before us, and they'll exist long after.

And they are beautiful

because we are lucky enough to be able to gaze upon them.

Their beauty is in our seeing.

When we approach the world with wonder,

it's not the answers that we find that ultimately give our lives meaning,

it is the questions.

The meaning is in our asking.

Because any technology,

no matter how exciting or novel or sophisticated,

is meaningless without the human beings at the center.

We, and we alone have the unique privilege to look with fresh eyes,

to ask the bigger question.

So look.

Ask.

Do not let go to waste the gift that we alone possess:

our capacity for wonder.

Thank you.

(Applause and cheers)