0:05

They had that like continuous Seinfeld

0:07

generator for a while. Oh

0:10

really? Yeah, I

0:12

forgot what happens. I

0:15

think it just went to and

0:17

started doing racist things though. It

0:19

started talking like Jerry Seinfeld. Yeah,

0:22

exactly. It started talking like Seinfeld

0:24

and Michael Richards. The

0:28

IDF is doing great things! Oh

0:31

lord. What's

0:33

with it? With these dude and

0:35

protesters. Well, that's the deal. They

0:37

all got the same tent! Who's giving them

0:39

the tents? That

0:42

is his material. Sound

0:50

is personal, intimate, and emotive. Just

0:52

like this podcast. We

0:55

are Audiosstack.ai. We

0:57

combine AI writing. The best synthetic

0:59

voices like ours. With

1:02

production like music and mastering. And

1:04

deliver them to be heard. Be it ads,

1:07

podcasts, or VOs for video. Just like this

1:09

ad you're listening to right now. However,

1:12

we have millions of spots just like this

1:14

on podcasts. And rather than hearing

1:16

from us, we want to hear from you. How

1:19

would you like to win an AI

1:21

audio campaign for free? Do

1:24

you work with businesses, products, events, or causes

1:27

that could benefit from free promotion on podcasts

1:29

in the coming month? Tell

1:31

us how you might use synthetic voices. Or

1:35

dynamically change ads for a news

1:37

podcast like this versus true crime,

1:39

history, or even comedy. Go

1:42

to audiosstack.ai/contest and

1:45

your company could be heard by millions. See

1:47

webpage for teas and sees. I'm

1:50

Katja Adler, host of The Global Story.

1:53

Over the last 25 years I've covered conflicts in the

1:55

Middle East, political and

1:57

economic crises in Europe, drug cartels, and other

1:59

countries. in Mexico. Now

2:01

I'm covering the stories behind the news

2:03

all over the world in conversation with

2:05

those who break it. Join me

2:08

Monday to Friday to find out what's

2:10

happening, why, and what it all means.

2:12

Follow the global story from the BBC

2:15

wherever you listen to podcasts. Here's

2:21

something you might not know about wireless.

2:23

Sometimes what you see isn't what you

2:26

get, but with visible what you

2:28

see is what you get. Switch to

2:30

visible the wireless company that makes wireless

2:32

visible. Get a one-line plan with unlimited

2:34

5G data powered by Verizon just $25

2:37

per month taxes and fees

2:40

included. Switch now at visible.com

2:42

monthly rate on the visible

2:44

plan for data management practices

2:46

and additional terms visit visible.com.

2:51

Hello the internet and welcome to

2:53

season 344 episode 2 of production

2:55

of iHeartRadio and

2:59

this is a podcast where we

3:02

take a deep dive into America's

3:04

shared consciousness America's deep brain if

3:07

you will a little tip of the cap because we're

3:09

big AI fans a little little spoiler we're

3:11

big AI fans now folks I have

3:14

seen the light come around I

3:17

think when it's all you see on

3:19

social media you're like this is art

3:21

oh yeah maybe this is cool yeah

3:24

it's Tuesday June 25th 2024 oh

3:28

yeah big day it's national

3:30

catfish day which is odd

3:32

because it's also my partner's

3:35

birthday I brag are you

3:37

have you been catfishing me this whole time

3:40

it's national strawberry haven't met her person still

3:42

right but one of these days no well

3:44

and every time I want a video chat

3:46

she says her phone's broken yeah we just

3:48

kind of stick to the phone call stuff

3:51

but it's also very normal at this stage

3:53

in a marriage very normal also this is

3:55

so weird and this is just like some

3:57

weird religious stuff it's National League did you

3:59

know this Does

8:00

it? Yeah, it comes through, washes the kibble, has

8:02

a few bites, and then takes off into the

8:04

night. It's like a pit stop for one of

8:07

the raccoons in the night. Yeah. I'm leading a

8:09

raccoon-based Dungeons

8:12

& Dragons campaign starting next week. Oh,

8:14

for real? That's going to be like,

8:17

yes, it's going to be like a heist that

8:19

takes place in the warehouse where

8:21

I play roller derby. I'm

8:24

very excited. I've got it really architected.

8:28

I don't want to give any secrets in

8:30

case any of my player characters listen to

8:32

this podcast. Right. Okay.

8:34

That sounds amazing. And what a coincidence. Raccoons,

8:36

having a bit of a moment. Yeah, truly.

8:38

At least on this podcast. Yeah, they are.

8:41

So in addition to being

8:43

a host of the wonderful

8:45

Mystery AI Hype Theater 3000 podcast, which

8:49

podcast hosts the highest honor one can attain

8:51

in American life, but you

8:53

both have some pretty impressive credits. Emily,

8:56

you are a linguist and professor at the

8:58

University of Washington, where you

9:00

are director of the Computational Linguistics

9:03

Laboratory. Yep, that's right. Do we

9:05

have that right? Okay. Alex, you

9:07

are director of research at the

9:10

Distributed AI Research Institute, both widely

9:12

published, both received a number

9:14

of academic awards, both have PhDs. We

9:16

had you on the podcast a few

9:19

months back, told everyone

9:21

the truth about AI that

9:23

a lot of the stuff that we're scared of

9:26

and a lot of the stuff we think it

9:28

can do is not true. It's bullshit. And

9:31

I sat back and was like, well, we'll

9:34

see what AI does after this one. And

9:38

it's just kept happening, you guys. What the

9:40

heck? What can we do? If

9:43

anything, it's gotten worse since we told

9:45

everybody the truth. What's happening? Truly.

9:48

You know, everybody seems to want

9:50

to believe. And it's absurd. Yeah. It

9:52

is so wild. Yeah. And

9:54

part of what we do with the podcast, actually, is

9:56

try to be a focal point for a community of

9:58

the people who are like... like, no,

10:01

that's not right. Why does everybody around me

10:03

seem to believe that it's like, you know,

10:05

actually doing all these things? Yeah. So

10:08

it's, yeah, it's, you know, it's

10:10

what we say in our podcast. Like every time

10:12

we think we've reached peak A.I. hype, the summit

10:14

of bullshit mountain, we discover there's worse to come.

10:16

Like it's not stopping. Yeah, right. Just keeps growing.

10:18

Oh yeah. This is just a base camp until

10:20

you get to the real peak. Like, right. Well,

10:22

it's just that you just keep on thinking that

10:24

keeps on becoming, and

10:27

there's more and more things that

10:29

the CEOs just, you know, are

10:31

really just say

10:33

incredible nonsense. I don't know if you saw

10:35

this. It was the

10:38

last week, the chief technology officer

10:41

of open A.I. Mira

10:43

Marati. Mirati. Yeah. Who

10:45

famously was, I

10:48

think it was an interview with 60 minutes. And

10:50

when they were talking about one of their tools,

10:52

Sora, you know, they had asked if they

10:55

ever filmmaker. Yeah. Exactly. Yeah.

10:58

Exactly. Reckon out Sora. Thank you.

11:00

Yeah, exactly. This saw really edging

11:02

out, you know, David Lynch these

11:04

days. And so, you know,

11:06

and they asked her, do you train the

11:08

stuff on YouTube and she she kind of

11:11

grimaced? Yeah. Painful. Yeah, we covered it. Yeah.

11:13

And and I remember a great Twitter comment.

11:15

I was like, well, if you're going to

11:17

just lie about stuff, you at least have

11:19

to have a good poker face about it.

11:21

Yes. And and so the last week she

11:24

was like, well, this is doing

11:26

another interview and she was like, well, some some some

11:29

creative jobs are going to go away.

11:31

Like some artists should be, you know,

11:33

please. Say some some creative jobs maybe

11:35

shouldn't have existed in the first place.

11:38

Right. Like these jobs were in a

11:40

front to God or something. Some of

11:42

them just shouldn't have even been there.

11:44

But she does have a French

11:46

accent. So it's really hard to be like,

11:48

this is ridiculous. Well, she's Italian. And

11:51

that's what's amazing about her having a friend. I

11:55

don't I'm not I'm not a cultured person.

11:57

I don't know the difference between. They're all

11:59

friends. to me, I'm American. Right.

12:02

Hey. She has a Canadian accent,

12:04

I think? I'm not sure. I know. It's

12:08

only plagiarism if it comes

12:10

from the French region of

12:12

Italy. That's right. That's

12:15

right. Yeah, we're going to get

12:17

into that story and just,

12:20

yeah. All of the madness

12:22

that has continued to happen, the bullshit

12:24

has continued to reign

12:26

even harder, it seems

12:29

like. Yeah. Which, yes,

12:31

does make the mountain go higher, unfortunately, the

12:33

bullshit mountain. But before we

12:35

get to that, Emily, Alex, we do

12:37

like to ask our guests, what is

12:39

something from your search histories that's revealing

12:42

about who you are? Alex, you want

12:44

to kick us off? Oh,

12:46

gosh. Okay. I don't... The

12:49

thing is, I don't think... So

12:51

I use DuckDuckGo, and so it doesn't actually

12:54

keep the search history. And

12:56

if I actually look at my Google history, it's

12:58

actually going to be really shameful. It's going to

13:00

be me searching my own name

13:02

to see if people are

13:04

shit-talking me online. No, this isn't just how

13:06

we tell if someone's honest, as if they

13:08

actually give that answer. We're like, okay, so

13:10

you are not a person. You actually search

13:13

yourself, yeah. But I think

13:15

the last thing I actually searched was queer

13:17

barbers in the Bay Area, because I haven't

13:20

had a haircut in like a year, and

13:22

I think I need to term up or

13:25

get air out the sides

13:28

of my head for Pride Month.

13:30

So that's the last thing I

13:32

searched. What

13:34

are you going? Are you going full shaved on

13:36

the sides? I think maybe trim it a little

13:38

bit, and trim it up the back, and bring

13:41

out the curls a little bit. Okay. Love

13:44

it. On board. I wish I could

13:46

bring out my curls. You've got a few more days in Pride Month

13:48

to get that done. I'm late. In

13:51

July, you're like, you do discounts? You

13:54

discounts? I'm late. It's like

13:56

after Valentine's Day. Do

13:58

I get an undercut at 50%? off now.

14:00

Right, exactly. Emily,

14:03

how about you? What's something from your search history? So

14:05

forgive the poor pronunciation of this and

14:08

the rest of the story because Spanish

14:10

is not one of my languages, but

14:12

champurado is something I search

14:14

into. Yeah. So I was in Mexico

14:16

City for a conference last week and at one

14:18

of the coffee breaks, they had coffee and decaf

14:20

coffee, and then they had champurado

14:22

con chocolaté o ajueña. And each of the

14:25

labels- You're kind of telling it

14:27

on the Spanish pronunciation by the way. Don't mean

14:30

to- Yeah, give us that about that. What do

14:32

you say when you see that word? Champurado Mexican

14:34

hot chocolate. All right.

14:36

So, yeah. You're literally reading the Google

14:38

results. Yeah. So

14:42

the labels all had like translations into English,

14:44

and so it was champurado with Oaxacan Choco.

14:46

I'm like, yeah, I got that. What's champurado?

14:49

And so I look it up because I want to know

14:51

what I'm consuming before I consume it. And it's basically a

14:54

corn flour based thick drink. So

14:57

like chocolate corn soup, it was

14:59

amazing. Chocolate corn soup. You

15:01

had me until chocolate corn soup.

15:06

The corn is just a thickening

15:08

egg there. Thick chocolate

15:10

drink. Yeah. Thick chocolate drink with

15:12

a slight corn flavor. Like think corn

15:14

tortilla, not crayon the cob. Yeah,

15:17

yeah. Ooh, yeah, yeah, yeah. That sounds amazing. Yeah.

15:20

It was really good. I love some corn flakes

15:22

in a chocolate bar. Uh-huh. Yeah. So corn and

15:24

chocolate. There you go. You got to arrive in

15:26

your own way as to why that appeals to

15:28

you. That's right. So I'm back on board with

15:31

the thick corn chocolate drink. It

15:34

was really good. And just awesome that it was

15:36

there. The coffee breaks had the Mexican

15:38

sweetbreads and stuff like that, but otherwise it was pretty

15:40

standard coffee break stuff. And then all of a sudden

15:42

there's this wonderful mystery drink. Yeah. One of the big

15:44

urns. It was lovely. That sounds great. What

15:46

is something you think is underrated, Emily? Um,

15:50

I think Seattle's weather is underrated.

15:53

Okay. Yeah. Everyone makes fun of

15:55

our weather and like, you know, fine. Believe that we don't

15:57

need lots of people coming here and it's true. It gets

15:59

dark. the winter, but like almost

16:01

any day you can be outside and you are

16:04

not in physical danger because you are outside. I

16:08

guess that's, I mean, if

16:10

you're going for, yeah, that's interesting. But I

16:12

mean, it's that, I mean, the winters are

16:14

just so punishing, though, it's so gray. It's

16:17

dark, but the weather is not going

16:20

to kill you. It

16:23

looks like shit, but experientially not

16:25

bad for you. I mean, yeah,

16:27

I know. It's, when does like,

16:30

it doesn't get all gloomy, I imagine in

16:33

the summer, right? You have wonderful blue skies

16:35

and you can enjoy the... The summers are

16:37

gorgeous. Yeah, summer's nice. Fire season aside. Right.

16:40

But yeah, from sort of mid-October to

16:42

early January, it can be pretty like,

16:44

it's gray. And so like when the

16:47

sun is technically above the horizon, it's a little hard to tell.

16:49

Yeah, right, right. So, but you know,

16:52

compared to like Chicago, where you have

16:54

maybe four livable weeks a year between

16:56

the too hot and the too cold.

16:59

Wow. Wow. Don't do that because my thing was going to

17:01

be Chicago because I was just there. And I

17:04

was going to say my answer was going to

17:06

be that Chicago is the

17:09

best American city. I

17:11

stand on this like 100%. For

17:15

two weeks out of the year, that's very true.

17:17

No, absolutely not true. No. Come

17:19

on. I'll even deal with the winter.

17:21

I'll deal with the winter. I mean,

17:23

if I... Okay, I'll

17:26

be honest. If I didn't, you know,

17:28

if the weather in Chicago, if I

17:30

could bring Bay Area weather to Chicago,

17:32

I would live in Chicago. I mean,

17:35

there's other reasons. But I mean, it's...

17:37

Look, the vibes immaculate.

17:40

Street festivals, the neighborhoods,

17:44

it's the one place that's probably... The food

17:47

still comparatively affordable compared

17:49

to the coast's radical

17:52

history. Just some of the

17:55

best politics. I would say... It's

17:59

not fugitive there. Yeah, they shot

18:01

what they shot. What did they shoot there?

18:03

The Fugitive? Oh, I did.

18:05

That's a deep cut. Yeah. I

18:07

mean, I think they've shot a lot of Batman

18:10

movies there because the iconic lower

18:12

Wacker Drive and

18:14

they call it Gotham. Yeah.

18:17

That's pretty cool. Yeah. Great city.

18:19

Crappy weather. If you're

18:21

going to dump on weather, I'm like, everyone makes fun

18:23

of Seattle's weather. Honestly,

18:26

Emily, this is a

18:28

hot take. I'd rather take Chicago's

18:31

weather than Seattle's weather because I

18:33

can't do gray. I can

18:35

do crossfire. I can do

18:37

frigid. I cannot do gray.

18:39

It's too depressing for me.

18:42

Well, this is why I say don't move to Seattle if

18:44

you can't handle our weather. The people who move here and

18:47

then complain about the weather are the worst. Yeah. It's

18:49

like, what'd you expect? Yeah. All of

18:51

this, what they say is true about it being gray and

18:53

they're like, oh, I didn't expect it to be that gray.

18:55

Right. I think people talk

18:57

about it like that. All

18:59

right, Alex, let's stay with you. What

19:01

is you guys's overrated and please do

19:04

it in a point counterpoint style also

19:06

that contradicts

19:08

one another. Well, I got to

19:10

think about what's overrated these days.

19:12

I just don't

19:14

know what's in the, I know the name of the

19:16

show is the Daily Zeitgeist, but I don't really know

19:19

what's in the Zeitgeist. I mean, I guess

19:21

Taylor Swift, I mean, I don't really

19:24

have, maybe that's controversial. I'm saying something

19:26

that's hot take, but I guess that's

19:28

maybe not controversial to people

19:30

of our

19:32

generation. No. So, yeah.

19:35

Joining Dave Grohl

19:38

on the attack this weekend. Yeah. Wait,

19:41

what happened with Dave Grohl? Dave

19:43

Grohl was implying that she's like,

19:45

he's like, well, we play our

19:47

music live, like raw live rock

19:49

and roll, you know, unlike the

19:51

errors tour, you know, the errors

19:53

tour. And then everyone's like, fuck

19:55

you, Dave. Or other people being

19:57

like, exactly, exactly. Yeah. It's

20:00

just like, yeah. Yeah. I mean,

20:02

Dave Grohl is also overrated, I

20:04

guess. But I mean, I enjoyed,

20:07

look, I enjoy Everlong, like the

20:09

next, like, yeah, middle-aged

20:12

sort of like dad figure. But

20:16

I, you know, I'm sure like

20:19

I'm glad that you played every part in

20:21

that song. It sounds good, but, you know,

20:23

yeah, it doesn't make you an

20:26

authority on Taylor Swift. So, yeah. So I

20:28

think I'm undercutting my own point. No, let's

20:30

go, Dave. You did Dave Grohl.

20:32

You got a point in your own overrated. In

20:34

my own, yeah. Which is excellent, because I don't

20:36

even have an opinion about Taylor Swift. Never saw

20:38

Tucker Carlson do that. Yeah. Was

20:41

that what that show was called, Crossfire? Or was that? Crossfire

20:45

was with, what's his face? Tucker

20:47

Carlson and Paul Bagala. That was

20:49

the one that John Stewart came

20:51

on and was like. Destroyed? Yeah.

20:54

It was like, this show is bad.

20:57

And then like they canceled it a couple of weeks

20:59

ago. But then there was

21:01

that one show, Hannity and Combs,

21:03

where Sean Hannity was supposed to

21:06

be conservative voice. And then, you

21:08

know, Combs were like, I don't even know

21:10

the guy's first name. They kind of

21:13

just had him as

21:15

a token, like liberal on. And then they

21:17

just, it was on Fox News that he

21:19

attacked him relentlessly. He wasn't allowed to read

21:21

the news. He's like,

21:23

you argue the liberal points, but you're actually

21:26

not permitted to leave this room. We're going

21:28

to keep you in here, old boy style

21:30

for the. Oh, that was the end of 60 minutes

21:32

that Andy Rooney would do. There

21:35

was part of 60 minutes was point counterpoint

21:37

and it would be Andy Rooney. If

21:40

that's what you're thinking, Jack. I don't know. There's

21:42

many. No, no, there was a show. Yeah. It

21:45

was right when I got out of

21:47

college and worked for ABC News. And

21:49

then at that time, there was a

21:51

big show on CNN called Crossfire. Yeah,

21:53

it was Tucker. But you are talking

21:55

about Crossfire. Yeah, Tucker Carlson was the

21:57

conservative. Paul Bagala was the liberal. And

22:00

they just like got on and yelled

22:02

at each other. I'm looking at now. This

22:05

is good. Apparently there there was a

22:07

that they they had

22:09

a revival. And then

22:11

in 2013 and 14 on the

22:13

left was Stephanie Cutter and Van

22:16

Jones and then

22:18

Newt Gingrich and S.E. Cup on

22:20

the right. And then

22:22

whatever. And then whenever they needed breaking

22:24

news, they'd bring in Wolf Blitzer for

22:26

some reason. Because

22:29

Wolf Blitzer. Attracting him out of the Situation

22:31

Room. Yeah. Yeah. Yeah. They released him from

22:33

the cryogenic. He was helicopter

22:35

lifted from the Situation Room, three

22:37

rooms over to the crossfire set,

22:40

just with dead pan. We need you. We

22:42

need you. No hint of emotion

22:44

on his face ever. You guys ever

22:46

seen the Wolf Blitzer episode

22:49

of Celebrity Jeopardy? No. Oh,

22:51

dude, you're so bad at it. A

22:53

favor. Is it? Is

22:55

it hasn't been scrubbed yet? Is

22:58

it as good as like the

23:00

SNL parodies, the Celebrity Jeopardy with

23:02

like the Sean Connery? It's just

23:04

so bad. Yeah. Just no,

23:06

no. And also

23:08

incorrect. Yeah. One

23:10

after another. Because he had like negative went

23:14

into the into the red. He's in the red.

23:16

Very quickly in Final Jeopardy. Well, Wolf, we're going

23:18

to spot you 3000 because we

23:20

can't have somebody be in negative

23:22

numbers going into going into Final

23:25

Jeopardy. And I think Andy

23:28

Richter was on with him

23:30

and just destroyed was so

23:33

good. That's so funny. Andy

23:35

Richter, like the

23:37

kind of crossover I didn't know I need. Yeah,

23:40

it's still up there mostly from what I could

23:42

tell. It's on YouTube. Yeah. I

23:44

am an old person. All

23:47

right. We still have Emily. You're

23:49

overrated. What do you think

23:51

is overrated? Big cars are overrated. Oh, totally.

23:54

Sort of halfheartedly looking for our next car

23:56

and can't find anything that is like reasonably

23:58

small. And the other day I was in

24:01

the parking lot for a grocery store near

24:03

here. Mostly I can walk for groceries, but

24:05

occasionally I have to drive to this other store. And

24:08

half the spots were labeled compact. And

24:11

all of those spots were taken up two at

24:13

a time by what we

24:15

now have as regular cars. Because somebody's

24:17

decided that people in this country don't

24:19

deserve normal size cars. Yeah.

24:22

There's so... I mean, it's to the point where

24:24

even the people who design parking lots are like,

24:26

we have to tell the automobile manufacturers. The

24:29

standard we've set as people who create

24:31

parking lots, they're pushing the boundaries of

24:33

what we can actually do or how

24:35

we measure things. Because the

24:37

cars are so fucking big. And

24:40

our streets around here in Seattle, we have a

24:42

lot of neighborhood streets where there's like parking on

24:44

both sides and then sort of just barely enough

24:46

space for two normal cars to go through. Right.

24:49

Or sometimes you have to pull over through the car and the bigger the car

24:51

is, the harder that gets. I love that thing.

24:53

I remember one of the times I went to

24:55

Seattle seeing how everybody just parks on whatever side

24:58

of the street in whatever direction they want. I

25:00

was like, all right. I'm like, all

25:02

right, Seattle. I was

25:04

not familiar. That's fun. Yeah. A

25:07

little bit of chaos. It totally offends my spouse

25:09

who's like, that's not how parking works. But

25:11

that's it. Yeah. I

25:14

love it. Yeah. Automakers

25:16

just seem to be getting bigger and heavier.

25:18

They won't stop until they make a car

25:20

that is legally required to have a fog

25:22

horn on it. Right. So

25:26

the Cybertruck? I was going to

25:28

ask, have you considered the Cybertruck?

25:30

I've seen one in person. They

25:32

are hilarious. Like you can't not laugh.

25:34

Exactly. It is

25:36

an experience seeing one in the wild. It's

25:38

like, wow. I just want to say

25:40

that what we really need is functional public transit. Yeah.

25:43

But short of that, we also need to not be doing bigger and

25:45

bigger cars. Yeah. Yeah. No,

25:48

I just, I mean, I have a truck. I have a 2020

25:51

truck and I wish, I really

25:53

wish it was much smaller because it's hard

25:56

to park. It's way too big.

25:58

I mean, I think the peak. of

26:00

truck design was like a 1987 Toyota

26:02

Tacoma long cab. Yes.

26:08

Just where like, yeah, you got to bunch up

26:10

your knees in the back if you want to

26:12

fit four people on it, but you

26:15

actually had a long, you actually had a

26:17

truck bed that actually had

26:20

some carrying capacity. And

26:24

it was a car you could absolutely run

26:26

into the ground with no problems at all.

26:28

Yeah. Oh yeah. My new

26:30

Ford Lightning needs a software update. Oh, God. Jesus

26:32

Christ. Well, that's the thing is like, yeah, I

26:34

mean, like, I mean, that's a big deal. I

26:37

know in like Oregon, which is like the way

26:39

they had right to repair bill. And

26:42

I mean, in some ways, the people that were

26:45

kind of into it weirdly were like, Google

26:47

actually came out kind of into it. There

26:49

was a good for four media

26:52

podcasts where they talked about this with

26:54

an Apple because they have such a

26:56

closed ecosystem was so against right to

26:58

repair. Even

27:00

if you have right to repair, they'd actually add

27:03

on all these things where you'd still have to

27:05

send to an authorized dealers because of firmware issues

27:07

or whatever. Right. Right. And

27:10

then John Deere, like John Deere is this kind of thing where they

27:12

have so much of their tractors

27:15

are computerized. And so there's like

27:17

a lot of like these John

27:19

Deere hacking kinds of things. Lots of people

27:22

who are outside of the US, you

27:24

know, programming these kinds of hacks for

27:26

people running these tractors and can't run

27:28

into their firmware. Yeah.

27:32

Yeah. The farmers have all the good, the

27:34

GPS, though something. But did you

27:36

hear about how the GPS was out for

27:38

a while with this? Oh, yeah. I

27:41

heard about that. This is again 404

27:43

media podcast, but the way those tractors

27:45

work for like planting is so precise.

27:47

Yeah. And the GPS. To

27:50

this end, I think. The GPS off, they basically

27:52

couldn't plant because then the seeds wouldn't be in

27:54

the right spot for the next process. Yeah. And

27:57

so they had to wait and there's a really

27:59

narrow window apparently with our, you know. currently genetically

28:01

modified, very, very specific corn that part of Monsanto

28:03

owns. And so it was actually looking pretty bad

28:06

for a while. I didn't hear any follow-up, so

28:08

maybe the solar flail was short enough and the

28:10

GPS came back online, but apparently that was a

28:12

big thing. Yeah. It

28:14

has to be a brief window because it goes

28:16

from corn seed planted

28:19

in the ground to popcorn in the

28:21

movie theater in two and a half

28:23

weeks. Yeah. It's a hot, hyper-engineered corn.

28:27

Most of it doesn't even go to popcorn in

28:29

the movie theater, most of it goes to animal

28:31

feed or ethanol, I think. Yeah. Right.

28:33

We're right. Yeah. All right. Well, let's

28:36

take a quick break and we're going

28:38

to come back and dive into why

28:40

Miles and I are excited

28:42

about the future of AI. We'll

28:45

be right back. Crossfire!

28:54

Zite gang, customers are rushing to

28:56

your store. Do you have a

28:58

point of sale system you can

29:00

trust or is it a real

29:02

POS? You need Shopify for retail.

29:04

Shopify POS is your command center

29:06

for your retail store. With Shopify,

29:08

you get a powerhouse selling partner

29:10

that effortlessly unites your in-person and

29:12

online sales into one source. Connect

29:15

with customers in line and online.

29:17

Get hardware that fits your business.

29:19

Take payments by smartphone. Transform your

29:21

tablet into a point of sale

29:23

system or use Shopify's POS Go mobile

29:25

device for a battle-tested solution. Plus, Shopify's award-winning

29:27

help is here to support your success every

29:29

step of the way. Do retail right with

29:31

Shopify. I was looking at their website just

29:34

trying to see, look, how do I find

29:36

the things that I need as somebody who

29:38

would potentially have a retail business? And surprisingly,

29:40

very easy to navigate. They have all the

29:42

things that you need answers to right there

29:45

on the website. So definitely check it out.

29:47

So sign up for a $1 per

29:50

month trial period at

29:52

shopify.com/TDZ, all lowercase. Go

29:54

to shopify.com slash TDZ to

29:56

take your retail business to the next

29:58

level today. shopify.com slash... Sound

30:02

is personal, intimate, and emotive. Just

30:05

like this podcast. We

30:07

are AudioStack.ai. We

30:10

combine AI writing. The best synthetic

30:12

voices, like ours. With production, like

30:15

music and mastering. And

30:17

deliver them to be heard, be it

30:19

ads, podcasts, or VOs for video. Just

30:22

like this ad you're listening to right now.

30:24

However, we have millions of spots just like

30:26

this on podcasts. And rather than hearing from

30:28

us, we want to hear from you. How

30:31

would you like to win an AI

30:33

audio campaign for free? Do

30:37

you work with businesses, products, events, or

30:39

causes that could benefit from free promotion

30:41

on podcasts in the coming month? Tell

30:44

us how you might use synthetic voices. Or

30:47

dynamically change ads for a news

30:49

podcast like this versus true crime,

30:51

history, or even comedy. Go

30:54

to AudioStack.ai/contest and your company

30:56

could be heard by millions.

30:59

See webpage for T's and C's. I'm

31:03

Katja Adler, host of The Global Story.

31:05

Over the last 25 years, I've covered

31:07

conflicts in the Middle East, political

31:10

and economic crises in Europe, drug

31:12

cartels in Mexico. Now

31:14

I'm covering the stories behind the news all

31:16

over the world in conversation with those who

31:18

break it. Join me Monday

31:20

to Friday to find out what's happening,

31:22

why, and what it all means. Follow

31:25

The Global Story from the BBC wherever

31:27

you listen to podcasts. And

31:38

we're back. We're back. So just

31:41

to, for people who haven't

31:44

listened to your previous appearance

31:46

in a while, I feel

31:48

like a broad overgeneralization, but it

31:51

feels like the stuff that AI is

31:53

actually being used for and capable of

31:56

is not what we're being told

31:58

about through the mainstream media. It

32:01

is not an autonomous intelligence that is

32:03

going to be the bad guy in

32:05

a Mission Impossible movie. I mean, it

32:08

is a bad guy in a Mission Impossible movie,

32:10

but it's not going to be a bad guy

32:12

in reality. In our relationship. Yeah. The

32:15

way that our actual president believes in this.

32:19

That was an amazing reveal that

32:21

Joe Biden basically watched Mission Impossible

32:23

and was like, we got to

32:25

worry about this AI stuff, Jack.

32:29

It's going to know my next move. But

32:31

it is like the

32:33

large language models are basically

32:35

more sophisticated, autocomplete. That is

32:38

telling you what its data

32:40

set indicates you want

32:42

to hear or what its data set indicates will

32:45

make you think it is thinking, talking like

32:47

a person. In many

32:49

cases, that means what they call hallucinating,

32:51

what is actually just making shit up.

32:54

What other jobs could you say that you're

32:56

like, sorry, that was hallucinating. Oh,

32:59

okay. Oh, good. You

33:03

wouldn't last long as a precog.

33:05

Yeah. I

33:08

would be the worst precog. On the

33:10

IRS size, hallucinating on that last tax return.

33:12

Can I get a do-over? Well,

33:15

can people talk about using this to do

33:17

your tax returns? Yeah, right. Yeah, there's actually,

33:20

I mean, in California, there's whatever

33:22

the Department of Tax and Revenue,

33:25

there was some great reporting on CalMatters

33:28

by Kari Johnson. He

33:30

was talking about how they were using

33:32

this thing, some language

33:35

model, to effectively

33:37

advise the agents

33:40

who respond to people who call into the California

33:43

franchise, and they're like, well, they're there and

33:45

they're like, well, the agents are still going

33:47

to have the last word. But

33:53

they're overworked. They're

33:55

going to read this stuff or meet them. Right,

33:59

exactly. Oh, you're going to use this as an

34:01

extra thing, just an extra

34:03

expense to do the product, do

34:05

your job even better? That doesn't

34:07

sound like a company necessarily. Yeah.

34:11

Yeah. So an interesting thing

34:13

that we're seeing happen, we pay

34:15

attention when there's an AI story

34:17

that captures the zeitgeist. We

34:20

covered the B minus version of a

34:22

George Carlin routine that came

34:24

out, they were like, AI just brought George

34:26

Carlin back from the dead. We

34:29

covered Amazon Fresh, having that

34:31

store where the cameras know

34:33

what you've taken. And so even if you

34:35

try and shoplift, like the cameras know, they're

34:37

going to catch it. And then you don't

34:39

even have to check out, you just walk

34:42

out and they like, it charges your account

34:44

because of AI. And then

34:46

what we're seeing is that

34:48

when the truth emerges, it

34:50

does not enter the zeitgeist. You

34:53

guys cover it on your show, which is why we're so

34:56

thrilled to have you back. But we

34:58

have updates on those two stories. Carlin, that

35:00

was just written by a person. The

35:04

Amazon Fresh, those videos were

35:06

being fed to people

35:08

working in India to try

35:11

to track where everything

35:13

was going, which was why there was like

35:15

a weird pause, like as

35:17

people were there like, oh, I

35:19

think we got, okay, yeah, we're just gonna

35:21

do a best guess. But

35:23

it's straight up like mechanical Turk. Which

35:27

again, Amazon named one of their

35:29

companies the Mechanical Turk.

35:31

So they know what's going

35:33

on. They knew what they were planning to do

35:36

here all along maybe. But is

35:38

that kind of the model you're seeing is

35:41

big flashy announcement. This is

35:43

what AI integration can do. And

35:45

then when it falls short, people just

35:47

kind of ignore it. Or

35:50

how does it seem from where you're sitting? Yeah,

35:53

we haven't seen really good

35:55

implosions yet. And surprising because

35:58

like the stuff that goes wrong goes like. really,

36:00

really wrong. And people are like, yeah,

36:02

well, it's just in its infancy, which

36:04

is a really, really annoying metaphor. Because

36:06

it first of all suggests that this

36:08

is something that is, like

36:10

a human, like an animal at least

36:13

that's a baby and can grow. It's

36:15

something that is learning over time. And

36:17

also sort of pulls on this

36:19

idea that we should be kind to these systems,

36:21

because they're just little babies, right? And so if

36:23

something goes wrong, it's like, well, no, that's just,

36:26

it's still learning. And we get all of these

36:28

appeals to the future, like how good it's going

36:30

to be in the future. And there is, at

36:32

this point, I think so much money sunk into

36:35

this, that people aren't ready

36:37

to like, let go and own up

36:39

to the fact that yeah, so, and

36:41

it is, I guess, too easy to

36:44

hire exploited workers for poor pay, usually

36:46

overseas to like, backstop the stuff. There's

36:48

also so you gave us the Amazon

36:50

Go stores actually being monitored by people

36:52

in India. There was one of the

36:55

self driving car companies admitted that their

36:57

cars were being supervised by workers in

37:00

Mexico. And remember the stats

37:02

on the yeah, yeah, it was it was it

37:04

was so Eric Voight, the

37:06

CEO of cruise. And

37:08

he had said, and then there was this reporting on

37:10

the New York Times, where

37:12

they said, you know, they use they use

37:14

humans. And then he was

37:16

like, well, wait, wait, wait, wait, you're, you're really blowing

37:19

out of proportion. We only use it something

37:22

of three to five percent of the time.

37:24

Like, that's a huge, huge

37:27

amount of out. And

37:29

he posted this himself on on Hacker

37:31

News, which is this, you know, kind

37:33

of like, I don't know, 4chan for

37:35

tech bros, I guess, well, I guess

37:37

for 4chan is 4chan for tech

37:39

bros. But I mean, it's, you know, but

37:42

like, with a little less overt racism, I

37:44

guess. Just a little. Yeah, just a slightly.

37:46

Yeah, it was still yeah, but we're seeing

37:48

this in a lot of different industries. At

37:50

the end of the day, it's just, this

37:53

is outsourcing humans. Janet Vertesi is

37:56

a sociologist at

37:58

Princeton. She has a a piece

38:00

in Tech Policy Press, which the

38:02

title is something like AI is

38:04

just forecasting, or is just outsourcing

38:07

2.0 effectively. Yeah,

38:09

we're seeing a lot of the same patterns that

38:11

we saw in the early

38:14

90s when these business process

38:16

outsourcing or BPA organizations were really

38:18

becoming all the rage in the

38:20

US. Right. The other thing

38:22

that I see a lot too is I felt

38:24

early on, especially when we were talking about it,

38:27

the thing that intrigued us was when everyone was

38:29

like, dude, this thing's going to fucking end the

38:31

world. It's how powerful AI is. I

38:34

have a whole plan to take

38:36

myself off this mortal plane if I have

38:38

to, the moment in which AI becomes sentient

38:41

and takes over. I

38:43

think it felt like maybe the markets were like, hey

38:45

man, you're scaring the kids, man. Do we have another

38:47

way to talk about this? I feel like recently I

38:49

see more of like, together when

38:52

we harness human intelligence with AI,

38:54

we can achieve a new level

38:56

of existence and ideation that has

38:58

not been seen ever in the

39:00

course of human history. I

39:03

saw that in the Netflix J-Lo

39:06

movie where the entire crux of

39:08

the film was this AI skeptic,

39:11

had to embrace the AI in order

39:13

to overcome the main problem, conflict in

39:15

the film, or just even now, like

39:18

with the CTO of OpenAI also doing

39:20

a similar thing when talking about how

39:22

AI, some creative jobs

39:24

are just going to vanish, but that's

39:26

because when the human mind harnesses the

39:28

power of the AI, we're going to

39:30

come up with such new things. That

39:33

feels like the new thing, which is

39:35

more like we got to embrace it

39:37

so we can evolve into this next

39:39

level of thinking, etc, computation or whatever.

39:41

You guys on, sorry, I was just going

39:43

to say Mystery AI, Hype Theater 3000 reads

39:46

the research papers so that we don't have

39:48

to, and Miles watches the J-Lo movies so

39:50

that you don't have to. I'm glad you're

39:52

watching the J-Lo because there's so

39:56

many different cultural touchstones of this. Yeah. I

39:58

had to look at it. because I thought

40:01

the movie you were talking about was the

40:03

autobiography, This

40:06

Is Me Now, a Love Story. I'm

40:08

like, there's a film? I was

40:10

like, there's an AI subplot in

40:12

that? Yeah. I didn't

40:15

know that JLo's life was a

40:17

complete cautionary tale

40:20

about AI and

40:22

the inevitability of it. Sorry,

40:26

Emily was about to say something. I just want to

40:28

be starchy. Our

40:31

colleagues, to meet Gabriel and Emil

40:33

Torres, coined this acronym TESCRIL, which

40:36

stands for a bundle of ideologies that are

40:39

all very closely related to each other. What's

40:41

interesting about the transition, you notice that they've

40:43

basically moved from one part

40:45

of the TESCRIL acronym to another. It's

40:48

all this stuff that's based on these

40:51

ideas of eugenics and really

40:53

disinterest in any actual

40:55

current humans in the

40:57

service of these imagined people

41:00

living as uploaded simulations in the far

41:02

long future. It's utilitarianism

41:04

made even more ridiculous by being taken

41:06

to an extreme endpoint. This thing like

41:09

it's going to kill us all comes

41:12

partially from the long-termism part of this, which

41:14

people are fixated this idea of we have

41:16

to, and it's ridiculous. They have

41:18

a specific number, which is 10 to the 58,

41:21

who are the future humans who are going to live as uploaded

41:24

simulations in computer systems

41:26

installed all over the galaxy. These

41:29

are people who clearly have never worked in IT support.

41:32

Because somehow the computers just keep running. Yeah, it'll

41:34

be fun. Yeah. The

41:36

idea is that if we don't make

41:39

sure that future comes about, then

41:42

we collectively are missing out on the happiness of those

41:44

10 to the 58 humans. That's such

41:46

a big number that it doesn't matter what happens now. I

41:49

always say when I relate the story that I wish I

41:51

were making this up, but there are

41:53

actually people who believe this. That's where the like, oh

41:55

no, it's going to

41:57

end us all stuff lives. before

54:38

at BetMGM. Signing up and playing is so

54:40

easy. Simply sign up using Code Buckeye and

54:43

receive up to $1,500 back in bonus bets

54:45

if you don't win your first bet. When

54:47

you register with BetMGM you can get instant

54:49

access to a variety of parlay selection features,

54:51

live betting options, and the best daily promotions

54:53

in the business. And with BetMGM at your

54:55

fingertips, every play and every game matter more

54:57

than ever. Place your money line, prop, and

54:59

parlay bets with a King of Sportsbooks today.

55:01

Sign up using Code Buckeye and receive up

55:03

to $1,500 back in bonus

55:06

bets if you don't win your first bet. That's

55:08

right! specific

1:02:00

technology is, but more someone who's like, learns

1:02:02

how to harness technology for this

1:02:05

other specific aim. Yeah. Yeah.

1:02:08

So surveillance is not synonymous with safety.

1:02:10

Like the one kind of one use

1:02:12

case for the word surveillance that I

1:02:14

think actually was pro public safety is

1:02:16

there is a study that long-term study

1:02:18

in Seattle called the Seattle Flu Study.

1:02:21

And they are doing what they call surveillance testing

1:02:23

for flu viruses. So they get volunteers to come

1:02:25

in and get swabbed and they are keeping track

1:02:27

of what viruses are circulating in our community. Right.

1:02:30

I'm all for surveilling the viruses. Yeah. Especially

1:02:32

if you keep the people out of it. Yeah. I

1:02:35

would add a wrinkle to that just because I

1:02:37

think that, I mean, there's a lot of surveillance,

1:02:39

I mean, that's the kind of technology, that's the

1:02:41

kind of terminology they use with health surveillance to

1:02:43

detect kind of virus rates and

1:02:45

whatnot. I would also add the

1:02:47

wrinkle that like a lot of those, you

1:02:49

know, organizations are really trusted by, distrusted by

1:02:51

marginalized people. Like what are you going to

1:02:53

do? What's it mean? Especially

1:02:56

thinking like, you know,

1:02:58

like lots of trans folks and

1:03:00

like, especially like under

1:03:02

housed or unhoused trans folks and just like, you're going

1:03:04

to do what? You want this data from me for

1:03:06

who? You know? Right. Yeah.

1:03:09

Yeah. Understandably. Especially because surveillance

1:03:11

in general, like, is not a safety

1:03:14

thing, right? It is maybe a like

1:03:17

safety for people within the walls of the

1:03:19

walled garden thing, but that's not safety, right?

1:03:22

That's, yeah. The other thing

1:03:24

about this is that what we call

1:03:26

AI these days is predicated on enormous

1:03:28

data collection. Right. And so

1:03:30

to one extent, it's just sort of an excuse to

1:03:32

go about claiming access to all that data. And

1:03:35

once you have access to all that data, you can do things

1:03:37

with it that have nothing to do with the large language models.

1:03:40

And so there is, you know, this is I

1:03:42

think less, typically less immediately like threatening to

1:03:44

life and limb than the applications that Alex

1:03:47

was starting with. But there's a lot of

1:03:49

stuff where it's like, actually,

1:03:51

we would be better off without all that

1:03:53

information about us being out there. And

1:03:56

there's an example that came up recently. So did you

1:03:58

see this thing about the system called Recolored? call that

1:04:01

came out with Windows 11. This

1:04:03

is such a mess. So

1:04:06

initially it was going to be by default turned on.

1:04:08

Oh yes. This is the

1:04:11

Adobe story too. Yeah. Yeah. Every five seconds it

1:04:13

takes a picture of your screen. Then

1:04:15

you can use that to like using AI, search

1:04:18

for stuff that you've sort of, and their example is something stupid. It's

1:04:20

like, yeah, I saw a recipe, but I don't remember where I saw

1:04:22

it. So you want to be able to search back through your activity.

1:04:25

Like zero thought to what this

1:04:27

means for people who are victims

1:04:29

of intimate partner violence. Right.

1:04:32

That they have this surveillance

1:04:34

going on in their computer that eventually

1:04:36

ended up being shipped as off by

1:04:38

default because the cybersecurity folks push back

1:04:41

really hard. By folks, I don't mean

1:04:43

the people at Microsoft, I mean the people out in the

1:04:45

world who saw this coming. Yeah. But that's another example of

1:04:47

like surveillance in the name of

1:04:49

AI that's supposed to be the sort of

1:04:52

helpful little thing for you, but like no thought to

1:04:54

what that means for people. It's like, yeah, we're just

1:04:57

going to turn this on by default because everybody wants

1:04:59

this obviously. Right. It's like, no, I

1:05:01

know how to look through my history actually.

1:05:03

I've developed that skill. I

1:05:06

don't need you to take snapshots of

1:05:08

my desktop every three seconds. But your

1:05:10

shows covered so many upsetting ways that,

1:05:14

it doesn't seem like it's people implementing

1:05:16

AI, it's companies implementing AI in a

1:05:18

lot of cases to do jobs

1:05:21

that it's not capable of doing. There's

1:05:24

been incorrect obituaries, you

1:05:26

know, I like, it's not even a big deal of breaking up

1:05:28

when we have a new distributor, it's a local du board provider,

1:05:40

but you know, just you know, I think we should them but

1:05:42

go back into what we were doing and go

1:05:45

back into that, exact

1:05:47

time around when you were there. And

1:05:50

it was good. No, Tizeko, I

1:05:52

mean, not so much. I think kilometres

1:05:54

who is in shifted has become planning AI in a lot

1:05:56

of cases to do jobs that it's not capable

1:06:00

of. doing. There's been

1:06:02

incorrect obituaries. Grok,

1:06:04

the Elon Musk one, the Twitter

1:06:06

one made up fake headlines about

1:06:09

Iran attacking Israel and public, put

1:06:11

them out as a major trending

1:06:13

story. You have this great anecdote

1:06:16

about a Facebook chatbot

1:06:18

AI responding to

1:06:20

someone had this very specific

1:06:22

question. They have a gifted

1:06:24

disabled child. They're like, does

1:06:26

anybody have experience with a

1:06:28

gifted disabled like to e

1:06:31

child with like this specific

1:06:33

New York public school program

1:06:35

and the chatbot responds. Yes, I

1:06:38

have experience with that and just like

1:06:40

made up because they knew that's what that's

1:06:42

what they wanted to hear. And fortunately,

1:06:44

it was like clearly labeled as an AI

1:06:46

chatbot. So the person was like, what, what

1:06:49

the black mirror? Yeah, but

1:06:52

World Health Organization, you know,

1:06:54

eating disorder institutions replacing therapists

1:06:56

with AI, like you

1:06:58

just have all these examples

1:07:03

of this going being used

1:07:05

where it shouldn't be and things

1:07:08

going badly. And like

1:07:11

there's a detail that I think

1:07:13

we talked about last time about Duolingo, where

1:07:18

the model where they let

1:07:20

AI take over some of the stuff

1:07:22

that like human teachers and translators were

1:07:24

doing before. And you made

1:07:27

the point that people who are learning the

1:07:29

language who are beginners are not in a

1:07:32

position to notice that the quality has dropped.

1:07:34

Yeah. And I

1:07:36

feel like that's what we're seeing

1:07:38

basically everywhere now is just the

1:07:40

internet is so big, they're just

1:07:42

using it so many different places

1:07:45

that it's hard to catch them

1:07:47

all. And then there's not an

1:07:49

appetite to report on

1:07:51

all the ways it's fucking up.

1:07:54

And so it just everything is

1:07:56

kind of getting slightly

1:07:59

too drastically. shittier at once.

1:08:03

And I don't know what to do with that. I

1:08:06

would say, yeah, well, go ahead, Emily.

1:08:09

What you do with that is you make fun of

1:08:11

it. That's one of our things, is ridiculous process to

1:08:14

try to keep the mood up, but also just show it

1:08:17

for how ridiculous it is. And

1:08:19

then the other thing is to really seek out

1:08:21

the good journalism on this topic, because so much

1:08:23

of it is either fake journalism output

1:08:25

by a large language model these days, or

1:08:28

journalists who are basically practicing access journalism, who are

1:08:31

doing the Jews thing, who are reproducing press releases.

1:08:33

And so finding the people who are doing really

1:08:35

good critical work and supporting them, I think is

1:08:37

super important. Yeah. But Alex, you were going to

1:08:39

say. Well, I was, well, you just teed me

1:08:41

up really well, because I was actually going to

1:08:43

say, you know, some of the people who

1:08:45

are doing some of the best work on it are like

1:08:48

four or four media. And

1:08:50

I want to give a shout out to them

1:08:52

because they're, you know, these folks are basically, you

1:08:55

know, they were at motherboard and

1:08:58

motherboard, you know, or

1:09:01

the whole vice empire

1:09:03

was basically, you know,

1:09:05

sunsetted. So they laid

1:09:07

off a bunch of people. So

1:09:10

they started this kind of journalist

1:09:12

owned and operated place. And, you

1:09:14

know, that focuses specifically on tech

1:09:16

and AI. And these folks

1:09:18

have been kind of in the

1:09:20

game for so long. They know

1:09:23

how to talk about this stuff without really

1:09:25

having this kind of being bowled

1:09:27

over, you know, there's

1:09:29

people who play that access journalism,

1:09:32

like Kara Swisher, who like kind

1:09:34

of poses herself as this person

1:09:36

who is very antagonistic. But

1:09:39

like, you know, right off the just like

1:09:41

fawning over like AI people. Yeah, like

1:09:43

all the time. Yeah, I trusted Elon

1:09:45

Musk and tell us like, well, why

1:09:47

did you trust this man in the

1:09:50

first place? Did you know

1:09:52

I was reading the Peter Thiel

1:09:55

biography, the contrarian and,

1:09:57

you know, and like it's

1:09:59

a very. It's a very harrowing read. I

1:10:02

mean, it's fascinating, but it was very harrowing.

1:10:04

It wasn't in the augo. It was pretty

1:10:06

like critical But like, you

1:10:09

know, they discuss the PayPal

1:10:11

days, you know 24 years ago When

1:10:14

you know Elon Musk was like, well,

1:10:16

I want to rename PayPal to X

1:10:19

And then and then everybody was like

1:10:21

why the fuck would you do that?

1:10:24

People are already using people are using

1:10:26

PayPal as a verb You

1:10:28

know this effectively the same thing you did

1:10:30

with Twitter like yeah, people are talking about

1:10:32

tweet as a verb Why would you say,

1:10:34

you know, just it's been like an

1:10:36

absolutely vapid human being with

1:10:39

no business sense anyways

1:10:42

That was a very long way of saying

1:10:44

cares so sure sets and then They'll

1:10:48

say also saying that there's lots of

1:10:50

folks. There's a number of

1:10:52

folks doing great stuff So I mean folks

1:10:54

at four or four Karen how who's independent?

1:10:57

But had been at the Atlantic and MIT

1:11:00

Tech Review and Wall Street Journal Carrie

1:11:02

Johnson who was at wired is

1:11:04

now at Cal matters. There's a lot of people

1:11:06

that really Report on AI

1:11:09

from the perspective of like the people

1:11:11

who it's honoring Rather than

1:11:13

starting from well, this tool can do X

1:11:15

Y and Z, right? You know, we really

1:11:17

should take these groups out their claims But

1:11:20

yeah, I mean the larger part of it

1:11:22

is I mean, there's just so much stuff

1:11:24

out there, you know And it's it's so

1:11:26

hard and it is like whack-a-mole and I

1:11:28

mean we're we're not journalists by

1:11:31

training I mean, we're sort of doing a

1:11:34

journalistic thing right now commentary

1:11:36

where I Think

1:11:38

we're I think I would not say we are

1:11:40

journalists. I always say we are doing a journalistic

1:11:43

thing We

1:11:48

are not doing original reporting sure sure but

1:11:50

it is Well, and you

1:11:52

know, I I would you know, I'm not I

1:11:54

don't know I'm not the I don't I don't

1:11:56

know who decides this is the court of journalism,

1:11:58

but you know reporting insofar as

1:12:01

looking at original papers and effectively

1:12:03

being like, okay, this is

1:12:05

marketing. This is why it's

1:12:07

marketing. Yeah, there's no there

1:12:09

there. Yeah. Rather than a

1:12:11

Wizzbang CNET article or something

1:12:14

that comes out

1:12:16

of a content mill and says,

1:12:18

Google just published this tool

1:12:20

that says you can find 18

1:12:24

million materials that are complete.

1:12:26

Well, it's like, okay, well, let's look at those claims

1:12:29

and upon what grounds do those

1:12:31

claims stand and how

1:12:33

that's a pretty poor thing.

1:12:35

I like to think of what we're doing is, first

1:12:38

of all, sharing our expertise in our specific

1:12:40

fields, but also modeling for people how to

1:12:42

be critical consumers of journalism. So

1:12:46

journalism adjacent, but yeah, definitely without training

1:12:48

in journalism. Yeah, yeah. But I think

1:12:51

we want to do the M&M article math, I

1:12:54

mean. Oh my gosh.

1:12:56

There's this article that has like

1:12:58

broken our brains because it just

1:13:00

has this series of sentences. That

1:13:03

I don't know that like, is everything is degrading

1:13:05

like journalism. There's that story about like the Daily

1:13:07

Mail was like Natalie Portman was hooked on cocaine

1:13:09

when she was at Harvard. You're like, no, that

1:13:12

was from that rap she did on SNL. And

1:13:14

that was like a bit, but because it gets

1:13:16

ingested. This thing's just great. And then the Daily

1:13:18

Mail had to be like, at the end they

1:13:20

corrected it. They're like, she was not. That was

1:13:22

obviously a satirical and that was due to human

1:13:25

error. Like they really leaned into that. You're like,

1:13:27

no, yeah, of course. Did I say by the

1:13:29

time that a fabricated quote of mine came out

1:13:31

of one of these things and was printed as

1:13:33

news? No. No. So I

1:13:36

also like Alex have searched my own name because I

1:13:38

talked to journalists and not that I like to see

1:13:40

what's happening. And I had, there was something in an

1:13:42

outfit called Bihar Prabha that attributed this quote to me,

1:13:44

which was not something I'd ever said. And

1:13:47

not anybody ever remembered talking to. So I

1:13:49

emailed the editor and I said, please take

1:13:51

down this fabricated quote and printer retraction because

1:13:53

I never said that. And they

1:13:55

did. So the article got updated, remove the thing

1:13:57

attributed to me. And then there was a. a

1:14:00

thing at the bottom saying we've attracted this, but

1:14:02

what they didn't put publicly, but he told me

1:14:04

over e-mail, that the whole thing came out of

1:14:06

Gemini. They

1:14:08

posted it as a news article. Of course. The

1:14:11

only reason I discovered it was it

1:14:13

was my own name, and I never

1:14:16

said that thing. Well, I need your

1:14:18

expertise here to decipher this Food and

1:14:20

Wine article that was talking about how

1:14:22

M&Ms was coming out with a pumpkin

1:14:24

pie flavored M&M, but very

1:14:27

early, normally pumpkin pie flavored things don't enter

1:14:29

the market till around August, like around when

1:14:31

fall comes. But M&Ms- This is why we

1:14:33

were covering it, because we are journalists. Yes,

1:14:35

we are journalists. We cover the important stories.

1:14:38

In May, pumpkin spice already?

1:14:40

No. But again, they

1:14:42

were saying this is because apparently

1:14:44

Gen Z and millennial consumers are celebrating

1:14:47

Halloween earlier, but this is this

1:14:49

one section that completely- Wait, wait, can

1:14:51

we back up? What? Yeah.

1:14:54

I don't know. That's what they're saying according

1:14:56

to their analysis. So

1:14:58

let me read this for you. Quote,

1:15:02

the pre-seasonal launch of the milk

1:15:04

chocolate pumpkin pie M&Ms is a

1:15:06

strategic move that taps into Mars

1:15:08

market research. This research indicates that

1:15:10

Gen Z and millennials plan to

1:15:12

celebrate Halloween by dressing up and

1:15:14

planning for the holiday about 6.8

1:15:16

weeks beforehand. Well,

1:15:18

6.8 weeks from Memorial

1:15:20

Day is the 4th of July. So you

1:15:22

still have plenty of time to latch onto

1:15:24

a pop culture trend and turn it into

1:15:26

a creative costume. I

1:15:29

don't- That's a chaos. It's all chaos.

1:15:31

That's a chaos, right? It doesn't make

1:15:33

any sense. I know. Look, wait. Wait,

1:15:36

I'm fixing this. I'm

1:15:38

fixating on 6.8. Exactly.

1:15:41

I'm fixating on two. What does that even mean?

1:15:43

What the fuck does that mean? And where did

1:15:45

Memorial Day come from in that? And what is

1:15:48

6.8 weeks for Memorial Day? Because

1:15:50

it's not any of the days that they said

1:15:52

it was. They said July 4th. And

1:15:55

also 6.8 weeks isn't

1:15:57

a real amount of time. That's 47. 6.6

1:16:00

days. What is even a 6.8 week? So

1:16:05

if this were real, it's

1:16:07

possible that they surveyed a bunch of people

1:16:09

and they said, when do you start planning

1:16:11

your Halloween costume? Those people gave dates and

1:16:13

then they averaged that. That's how you could

1:16:15

get to it. I get that. That's

1:16:17

fair. But also, it totally

1:16:19

sounds like someone put into a large

1:16:21

language model, write an

1:16:23

article about why millennials and Gen

1:16:25

Z are planning their

1:16:28

Halloween costumes earlier. It sounds

1:16:30

like that. But also just so odd

1:16:32

to say, well, 6.8 weeks from Memorial

1:16:34

Day is the 4th of July. This

1:16:36

article didn't even come out. It came

1:16:38

out after Memorial Day. It's

1:16:41

just nothing made sense. I was like,

1:16:43

I don't fucking understand what they're doing

1:16:45

to me right now. But again, this

1:16:48

is the insidious part for me about it. So

1:16:50

this appeared in Food and Wine? This is in

1:16:52

Food and Wine magazine with a human in

1:16:55

the byline. I actually DM'd this person on

1:16:57

Instagram, and I said, do you mind just

1:17:00

clarifying this part? I'm a little bit confused

1:17:02

and I've got no response.

1:17:05

I'm wondering if it's because I know

1:17:07

that there was some good coverage in

1:17:09

Futurism, and they were

1:17:12

talking about this company called Advon

1:17:14

Commerce, and the way that

1:17:16

basically this company has been basically

1:17:19

making AI generate

1:17:22

articles for a lot of

1:17:24

different publications, usually on

1:17:26

product placement. So

1:17:30

it makes me think it's like, because

1:17:32

Food and Wine may have

1:17:34

been one of their, I forgot the article,

1:17:37

but they had better homes and

1:17:40

gardening and these legacy articles like

1:17:42

that. So I don't know if

1:17:44

it's something of that or

1:17:46

this journalist said, write me this

1:17:48

thing and I'm just going to drop it and then

1:17:50

go with God. Yeah.

1:17:55

My other favorite example of AI is this

1:17:57

headline I saw somewhere. It's no big secret.

1:18:00

why Van Vaught isn't around anymore. And with

1:18:02

a picture of Vince Vaughn, but they just

1:18:04

got his name completely

1:18:06

wrong. Yeah. Well, I can't find it.

1:18:10

It's no big secret. Why

1:18:13

Van Vaught isn't around anymore. You

1:18:19

know, if I was just scrolling and I'd

1:18:22

say, yeah, I liked Van Vaught

1:18:24

and the intern. And

1:18:29

then I would have looked at it and then I

1:18:31

would have double taped. I'm like, wait, wait, wait. Did

1:18:33

he co-star with Owen McWilson

1:18:35

or something? Yeah, yeah, yeah, yeah. Russell

1:18:38

Wilson was in that. I

1:18:40

think it was the Adweek report that you're

1:18:42

thinking about. So Futurism did a bunch of

1:18:44

it, but then Adweek had the whole thing

1:18:47

about Advon and I can't quite get through

1:18:49

it. No, it was Futurism. It was Futurism,

1:18:51

yeah. Because Adweek had the thing on this

1:18:53

program that Google was offering

1:18:55

and it didn't have a name. Oh, right. Yeah.

1:18:57

So Advon was Futurism. Yeah, but it totally sounds

1:18:59

like one of those. But it is

1:19:01

happening, yeah. Yeah. See, I thought you were going

1:19:03

to talk about the surveillance by M&M thing. We

1:19:06

said M&Ms. So this was somewhere

1:19:08

in Canada. There was an M&M vending machine

1:19:10

that was taking pictures of the students while

1:19:12

they were making their purchases. And I forget

1:19:14

what the sensible purpose was, but

1:19:17

the students found out and got

1:19:19

it removed. Probably freaked out

1:19:21

and made a big deal about it. Students,

1:19:24

are we right? Well,

1:19:27

I feel like we could talk to you guys

1:19:29

once again for three hours. There's

1:19:32

so much interesting stuff to talk about.

1:19:34

Your show is so great. Thank you

1:19:36

both for joining. Where can

1:19:38

people find you, follow you,

1:19:40

all that good stuff. Emily, we'll start

1:19:42

with you. Well, first, there's

1:19:44

the podcast, Mystery AI Hype Theater 3000,

1:19:46

where you find any podcast, you can

1:19:49

find ours. And we've also started a

1:19:51

newsletter. If you just search Mystery

1:19:53

AI Hype Theater 3000 newsletter, I think it'll turn

1:19:55

up. And that's an irregular newsletter,

1:19:57

where we basically took the things that used

1:20:00

to... be sort of little tweet storms. And

1:20:02

since the social media stuff has gotten

1:20:05

fragmented, we're now creating newsletter posts with

1:20:07

them. So it's off the cuff discussions

1:20:09

of things. On Twitter,

1:20:12

X and Macedon and Blue

1:20:15

Sky, I'm Emily M Bender. And

1:20:17

I'm also reluctantly using LinkedIn

1:20:19

as social media these days. So the news

1:20:22

that I need is... It's gonna

1:20:24

be the last one. It's gonna be the one

1:20:26

that survives them all because we... I know. ...some

1:20:28

people kinda need it. Really the cockroaches of social

1:20:30

media on the website. Yeah.

1:20:33

Yeah. Yeah. Yeah. Yeah. I'm

1:20:35

at Alex. Oh, you Alex? Yeah.

1:20:37

Alex Hanna, H-A-N-N-A on

1:20:39

Twitter, Blue Sky. I barely

1:20:43

use Blue Sky or Macedon, but Twitter is

1:20:45

the best place to find me. Also

1:20:48

check out dare, dare,

1:20:51

D-A-I-R, hyphen, institute.org. We're

1:20:54

also dare, underscore, institute on Twitter,

1:20:57

Macedon. And we're not on

1:20:59

Blue Sky yet, but we're

1:21:02

on LinkedIn. But that's where

1:21:04

you learn a lot about what our institute's doing, lots

1:21:08

of good stuff, amazing colleagues, and

1:21:11

whatnot. Yeah. Amazing. And

1:21:13

is there a work of media

1:21:15

that you've been enjoying? Yes.

1:21:18

I've got one for you. This, I think, started off

1:21:20

as a tweet, but I saw it as a screencap

1:21:22

on Macedon. So it's by Lama in a text. And

1:21:24

the text is, don't you understand that the human

1:21:26

race is an endless number of monkeys. And every

1:21:29

day we produce an endless number of words. And

1:21:31

one of us already wrote Hamlet. That's

1:21:35

really good. That's such

1:21:38

a hyper-specific piece of media.

1:21:42

I think last time I was on this,

1:21:45

I was plugging Worlds Beyond Number, which is

1:21:47

a podcast, which I'm just absolutely in love

1:21:49

with, which is a Dungeons

1:21:51

and Dragons actual play podcast. But it's

1:21:53

got amazing sound production. I

1:21:56

would just plug in everything on

1:21:58

dropout.tv. I mean, it's a streaming

1:22:00

service, honestly. Sam Reich, who is

1:22:03

the son of Robert Reich,

1:22:05

liberal darling and

1:22:10

former Department of Labor Secretary in

1:22:13

the Clinton administration, has

1:22:15

turned college humor into

1:22:17

an area of really

1:22:19

great comedians. So

1:22:22

they're putting out a lot of great stuff.

1:22:24

So I'd say make

1:22:26

some noise. It's coming out with a new season today,

1:22:28

which is a really great improv

1:22:30

comedy thing. And yeah,

1:22:33

let's just go with that. So just

1:22:35

plug in. Those very important people interviews

1:22:38

are hilarious. Those very important interviews, Vic

1:22:40

Michaelis. I named one of

1:22:42

my chickens, vehicular manslaughter, after

1:22:45

an inside joke there, and another one,

1:22:48

Thomas Shrigley. So yeah,

1:22:50

just incredible, incredible stuff. Yeah.

1:22:53

Shout out to Sam. He's one of the

1:22:55

best. Miles. Yes. Where

1:22:57

can people find you? Is there a

1:22:59

work media you can enjoy? They have

1:23:01

at symbols. Look for at Miles of

1:23:03

Gray. I'm probably there. You

1:23:05

can find Jack and I on our

1:23:08

basketball podcast, Miles and Jack at Mad

1:23:10

Max D's, where we've wrapped up the

1:23:12

NBA season and I have two streaming

1:23:15

down my face with pain and anger

1:23:17

as the Celtics win again. And also

1:23:19

if you want to hear me talk

1:23:21

about very serious stuff, I'm talking about

1:23:23

90 day fiance on my other show,

1:23:25

420 day fiance, which you can check

1:23:28

out wherever they have podcasts. A tweet

1:23:30

I like first

1:23:32

one is from a past

1:23:34

guest, Josh Gondelman, tweeted, I

1:23:37

bet the best part of being in a

1:23:39

throuple is he have someone to do all

1:23:42

three Beastie boys, Parsa karaoke. I

1:23:45

guess one way to look at that. And

1:23:47

then another one from other past guest Demi

1:23:49

Adijuibe at electro lemon, uh, got his account

1:23:51

hacked and he tweeted, hi, hello, it's Demi.

1:23:53

I got my account back. Uh, I feel

1:23:55

the need to clarify that under no circumstances

1:23:57

should you ever believe that. or

1:24:00

anybody on this website is selling cheap

1:24:02

Mac books for charity or otherwise. And

1:24:04

what benefit would my signature do to

1:24:07

a laptop? So yeah, thank you for

1:24:09

clarifying it. I actually remember because I

1:24:11

followed Demi and I remember when his

1:24:13

account got hacked and I thought, man,

1:24:16

that's really, and I, at first I

1:24:18

thought it was a bit because Demi

1:24:20

is hilarious. But then I'm just like,

1:24:22

what the hell? It's funny, his follow-up

1:24:24

tweet was, for anyone who

1:24:26

thought I was doing a bit, what's

1:24:28

the punchline? My

1:24:31

jokes are never so obtuse. I love it

1:24:33

when you pay off. I want you to

1:24:35

know it wasn't all that funny and I

1:24:37

want you to know quick. Yeah, no, I

1:24:39

was also trying to find out what the

1:24:41

punchline was. Right, right. Yeah. Wait for it.

1:24:43

Wait for it. He's so funny

1:24:45

that part of you wants to be like, well, hold

1:24:47

on. What are you doing here? Yeah, what's the deal

1:24:49

here? You don't want

1:24:52

to immediately just dismiss Demi because he's

1:24:54

such a great comedic. Yeah. But yeah,

1:24:56

if you do want good Demi content,

1:24:58

the who's welcome at the cookout, you

1:25:00

can find that some dropout content that you can

1:25:02

get for free on YouTube. There you

1:25:04

go. We have been enjoying

1:25:07

Sleepy at Sleepy underscore nice tweeted.

1:25:09

It's absurd that Diddy Kong wears

1:25:11

a hat that says Nintendo, patently

1:25:13

ridiculous. There's no way he understands

1:25:16

the significance. It would be like

1:25:18

me unknowingly wearing a hat that

1:25:20

coincidentally depicts the true form of

1:25:22

the universe. That's

1:25:27

incredible. Oh my

1:25:29

God. It's

1:25:32

so fucking good because yeah, the second he

1:25:34

showed up, you're like, I don't know. Yeah,

1:25:36

I know. Brandon, he likes

1:25:39

Nintendo. You can find

1:25:41

me on Twitter at Jack underscore O'Brien.

1:25:43

You can find us on Twitter at

1:25:45

daily zeitgeist. We're at the daily zeitgeist

1:25:47

on Instagram. We have a Facebook fan

1:25:49

page and a website daily zeitgeist.com

1:25:51

where we post our episodes and our

1:25:53

foot. No, we link off to the

1:25:55

information that we talked about in today's

1:25:57

episode as well as a. that we

1:26:00

think you might enjoy. Miles,

1:26:02

what song do you think people might enjoy?

1:26:30

What does it actually come out? Is it recent? Hey

1:27:01

there girls, where are you going?

1:27:03

And they're like, down to the

1:27:05

beach is where we're going. But

1:27:10

there's this like charm to it and the

1:27:13

instrumentation is cool. So anyway, this is the

1:27:15

beach nuts with out in the sun parenthetical

1:27:17

fail. All right. Well,

1:27:19

we will link off to that in the footnotes. The

1:27:21

daily zeitgeist is a production of I

1:27:23

heart radio for more podcasts from my heart radio

1:27:26

visit. Yeah. Heart radio app, Apple podcast, or

1:27:28

wherever a fine podcast are given away for free. That's

1:27:30

going to do it for us this morning.

1:27:32

We're back this afternoon to tell you what

1:27:34

is trending and we will talk to y'all

1:27:37

then. Bye. Bye. Bye. Sound

1:27:45

is personal, intimate and emotive. Just

1:27:47

like this podcast, we

1:27:50

are audio stack.ai. We

1:27:52

combine AI writing the best synthetic

1:27:54

voices like ours with production

1:27:57

like music and mastering and

1:27:59

deliver the them to be heard, be it

1:28:01

ads, podcasts, or VOs for video.

1:28:03

Just like this ad you're listening to right now.

1:28:06

However, we have millions of spots just

1:28:08

like this on podcasts. And rather

1:28:10

than hearing from us, we want to hear from you.

1:28:13

How would you like to win an

1:28:15

AI audio campaign for free? Do

1:28:19

you work with businesses, products, events, or

1:28:21

causes that could benefit from free promotion

1:28:23

on podcasts in the coming month? Tell

1:28:26

us how you might use synthetic voices. Or

1:28:30

dynamically change ads for a news

1:28:32

podcast like this versus true crime,

1:28:34

history, or even comedy. Go

1:28:36

to audiostack.ai/contest, and your company

1:28:39

could be heard by millions.

1:28:41

See web page for Ts and Cs. Here's

1:28:44

something you might not know about

1:28:46

wireless. Sometimes what you see isn't

1:28:48

what you get. But with Visible,

1:28:51

what you see is what you

1:28:53

get. Switch to Visible, the wireless

1:28:55

company that makes wireless visible. Get

1:28:57

a one-line plan with unlimited 5G

1:28:59

data powered by Verizon. Just $25

1:29:01

per month, taxes and fees included.

1:29:03

Switch now at visible.com. Monthly

1:29:06

rate on the Visible plan. For

1:29:08

data management practices and additional terms,

1:29:10

visit visible.com.