0:00

Hey Grown Ups, this episode of Wow in the

0:02

World is brought to you by Huggy's Little Movers.

0:05

Huggy's knows that babies come in

0:07

all shapes and sizes, and parents

0:09

know that there's nothing worse than

0:11

an ill-fitting diaper, especially on an

0:13

active baby. Phew!

0:15

Huggy's Little Movers offer 12-hour protection against

0:17

leaks, and are curved to fit all

0:20

curves. So your babies will feel comfy

0:22

no matter how much they move around.

0:24

And they do tend to move around. A lot.

0:27

Get your baby's butt into the best-fitting

0:29

diaper. Huggy's Little Movers. We got you,

0:31

baby. Huggy's

0:57

Little Movers. Huggy's

1:10

Little Movers. Huggy's

1:15

Little Movers. Hey,

1:18

yeah, it's summertime. No

1:24

homework, no school. Yeah,

1:29

it's the hottest season. Me

1:35

and Reggie are keeping it cool. Sku's

1:43

out, sun's out, time for fun. Hanging

1:51

with my friends and

1:53

everyone. I'm

2:00

good. I'm good. I'm good. Two,

2:07

three, four! We

2:11

will summon! Summon! Summon! Forever

2:14

and forever! We will

2:16

summon! Show

2:19

them the reflection and do cool stuff,

2:21

yeah! We will summon! I

2:24

don't know what kind of stuff, but we'll figure it out!

2:27

We will

2:29

summon, yeah!

2:35

Hello and welcome to We

2:37

Will Summer Week! Summer

2:39

fun forever and ever! That's

2:42

right, Reggie. This is the week where we have

2:44

all kinds of summer fun, but we couldn't think

2:46

of anything to do, so

2:48

we're going around the neighborhood and bother-y,

2:50

I mean, getting inspiration from everyone else.

2:54

Are you ready to go find someone to talk to, Reggie?

2:57

Okay, let's go! Wack,

2:59

wack, wack, wack, wack! Snooping,

3:02

snooping, snooping! Oh

3:06

hey, look, it's Grandma G for you! Hey,

3:09

Granny G! Where did

3:11

you come from, boy? What do you want?

3:13

Well, I'm just out having summer fun forever

3:15

and ever and wanted to check in on

3:18

my favorite, Mindy's grandma. Well,

3:21

I'm just busy painting this

3:23

diaper house still. Oh

3:26

wow, it looks great! And it's

3:28

gonna stay that way. Now

3:31

get away now, get back! What

3:33

are you painting? Some kind of

3:35

mural? Yeah, it's me and my

3:37

electric guitar riding a saber-toothed tiger.

3:39

Oh, come on! Can

3:41

I paint a picture of me on there too?

3:43

No. Can I paint

3:46

sunglasses on the tiger? No. Can

3:48

I paint calligraphy flourishes and whirls around the

3:50

edges? No. Well then,

3:52

what can I do? Well,

3:55

you can wash my brushes.

3:57

Here, take them. Hooray! For

16:00

bananas go in the

16:02

yellow category. But

16:05

it sorts these pictures into

16:07

hundreds of different categories, like

16:10

their shape, their size, their

16:12

shininess, their brightness. The banana

16:14

also goes in the long

16:16

category and also also goes

16:19

in the medium sized category

16:21

and the non-shiny category and

16:23

the fruit category. So at

16:26

the end of this class,

16:28

the AI knows both what

16:30

these images are. That is

16:32

the ball. But also the

16:35

characteristics are things that make

16:37

up that image. The ball

16:39

is white. It

16:41

has stitches. It

16:43

is round. It is

16:45

used in the game of baseball.

16:48

Wonderful. You're now ready

16:51

for semester two, creation. Yay.

16:54

I finally get to make my

16:56

own art. So now when

16:58

you type a prompt into the machine,

17:01

ahem, one picture of an

17:03

ice cream, please. Hmm,

17:05

okay. Ice cream,

17:08

round, colorful, moist,

17:11

shiny. It

17:13

then arranges a bunch of

17:15

pixels into a shape that

17:17

has those characteristics, like shape,

17:19

quality and shininess. And pixels

17:21

are the tiny dots on

17:23

a screen that make up

17:25

an image. Exactly. Hundreds

17:27

of thousands of these little colored

17:30

dots that all come together to

17:32

make a picture. And the AI

17:34

shuffles these around based on what

17:36

it knows about the object you

17:39

asked it to generate for you.

17:41

Exactoritos. Duh.

17:44

What do you think, Mrs. Teacher

17:46

Man? It's horrible. Oh

17:48

no. The first attempts are never

17:51

good, but then the AI

17:53

goes away and improves it, moving the

17:55

pixels around just a little bit. And

17:57

it does this over and over and

18:00

over. again thousands of times a second.

18:03

How's this? No. This?

18:07

No. This? No.

18:10

This happens over and

18:12

over and over again,

18:14

improving little by little

18:16

until... This? Oh,

18:19

now that's what I call an A

18:21

plus ice cream. Excellent.

18:24

Here you are, man, with just one

18:26

picture of ice cream. I love it!

18:32

And that, Guy Ross, is how

18:34

you train an AI generator to

18:36

become an artist. Oh, wow! That

18:39

was quite a lot to take in, Mindy. So

18:42

let me get this straight. This

18:44

AI program is taught what images

18:46

look like. Uh-huh. And then it's

18:48

taught the characteristics of those images,

18:51

like its size and color and

18:53

shape. Uh-huh. And

18:55

then it puts a bunch of

18:57

pixels together that have those characteristics.

19:00

You know it. And then the

19:03

AI keeps rearranging those pixels over

19:05

and over again, making slightly better

19:07

images every time until it has

19:10

a brand new picture of whatever

19:12

you asked for. Yep. That's

19:15

it in a nutshell. Wow. That's

19:17

kind of incredible. Yeah, almost

19:19

as incredible as Reggie

19:21

riding a skateboard over

19:24

the Grand Canyon. Check

19:29

it out. Wow. Okay,

19:31

what should I put in next? How about...

19:34

I've got it! How

19:36

about three best friends? Uh,

19:38

sure. Yay! Three

19:40

best friends! Three best friends! Three best

19:42

friends! But, Mindy, aren't you worried about

19:44

these machines replacing artists in the real

19:47

world? I mean, it might take a

19:49

real human artist months or even years

19:51

to create a piece of art like

19:53

this. But these machines

19:55

can do it in the blink

19:57

of an eye. Well, that's a

19:59

good point, Gyroz. some people are

20:01

worried about these machines. I mean,

20:03

on one hand, you can see

20:05

them as just another tool for

20:08

artists to express themselves. Right? Because

20:10

these machines aren't creative. To them,

20:12

these images are just a bunch

20:14

of pixels and numbers. It's us

20:16

humans who input the prompts to

20:18

make them generate art. Huh. But

20:20

on the other hand, some people are

20:22

worried about these AI generators copying real

20:25

life artists and not giving them credit

20:27

for their work. What

20:29

do you mean? Well, remember how

20:31

that AI generator had to go

20:33

to art school to learn how

20:35

to generate? Yeah, the University of

20:37

the World Wide Web. Right. Well,

20:39

when it's learning how to recognize

20:41

and sort objects, it's doing that

20:43

based on millions and millions of

20:45

images from the web. Uh-huh. And

20:47

a lot of those images are

20:49

created by artists who never gave

20:51

their permission for their art to

20:53

be used to teach robots. Right.

20:55

And I'm guessing that when these

20:57

AI art generators are asked to

20:59

produce a work of art, it

21:01

can sometimes end up looking very

21:03

similar to those artists it was

21:05

trained on. Exactoritos. So is this

21:07

a good thing or a bad

21:09

thing? Well, it's a new thing,

21:11

Gyros. It's up to us humans

21:13

to make that choice of how

21:15

we use it. It's done!

21:18

It's done! The three best

21:20

friends! Look! All right. Oh,

21:22

look, there's Mindy. And that's

21:25

Guy with his glasses. But

21:28

wait a minute. Mindy, I think this

21:30

machine is broken. Oh, boy. This one

21:32

in the middle doesn't look anything like

21:34

me. It's got a tail and feathers

21:36

and a beak. Uh. Ah,

21:40

Rachi! Ha ha ha! Grown-ups,

21:47

if you like Wow in the

21:49

World, you can listen early and

21:51

ad-free right now on Wondery Plus.

21:53

Join Wondery Plus in the Wondery

21:56

app or on Wondery Kids Plus

21:58

on Apple Podcasts. Prime members can...

22:00

and listen ad-free on Amazon Music.

22:02

And before you go, tell us

22:04

about yourself by filling out a

22:06

short survey at wondery.com/survey. Wow

22:12

in the World is written by

22:14

Mindy Thomas and Tom Van Calken

22:16

with help from me, Guy Roz.

22:19

Original sound design and music editing

22:21

is done by Tyler Thole with

22:23

help from our supervising producer, Jed

22:26

Anderson. You can also hear Jed

22:28

Anderson in the voices of Dennis,

22:30

Thomas Fingerling, Reggie, and many of

22:32

the other silly characters you hear

22:35

on our show. Jessica Bodie keeps

22:37

our facts straight as our fact

22:39

checker and Meredith Halpern-Ranzer powers the

22:42

wow at Tinkercast. Our theme song

22:44

was composed and performed by three-time

22:46

Grammy nominees, The Pop-Ups. Find them

22:49

at thepopups.com. Special thanks

22:51

to Steph Sosa, Henry Moskle,

22:53

Ruth Morrison, Rebecca Caban, Kit

22:55

Ballinger, Sana Ali-Mohamed, Anna Zegorski,

22:57

and all of the other

22:59

Tinkerers at Tinkercast HQ. To

23:01

keep the wows rolling, visit

23:03

us at tinkercast.com. There

23:06

you can learn more about becoming

23:08

an official member of the World

23:11

Organization of Wowsers. Learn about upcoming

23:13

events, shop our wow shop, find

23:16

our best-selling books, and learn about

23:18

all of our other amazing podcasts

23:20

from Tinkercast. Grownups, you can follow

23:23

Wow in the World on Facebook,

23:25

Instagram, and Twitter at

23:27

wowintheworld, and our email

23:30

address is hello at

23:32

tinkercast.com. And if

23:34

you're a kid with a big wow to share

23:36

with us, call us at 1-888-7-WOW-WOW for

23:40

a chance to be featured at the

23:42

end of our show or an upcoming

23:44

episode of Two What's in a Wow.

23:46

Thanks again for listening, and until next

23:48

time, keep on wowing! Wow

23:50

in the world! Wow in

23:53

the world! Wow in the world!

23:55

Wow in the world! Wow in

23:57

the world was made by Tinkercast.