0:00

In a world of economic uncertainty

0:02

and workplace transformation, learn to

0:05

lead by example from visionary

0:07

C-sweet executives like Shannon Schuyler

0:09

of PWC and Will Pearson

0:12

of Ihart Media. The good

0:14

teacher explains, the great teacher

0:16

inspires. Don't always leave your team

0:18

to do the work. That's been

0:20

the most important part of how

0:22

to lead by example. Listen

0:24

to Leading by Example, executives making

0:27

an impact on the I Heart

0:29

Radio app, Apple Podcasts, or wherever

0:31

you get your podcasts. Hi,

0:36

I'm Morgan Sun, host of Close All

0:38

Tabs from KQPD, where every week

0:40

we reveal how the online world

0:42

collides with everyday life. You don't know

0:44

what's true or not, because you don't know

0:47

if AI was involved in it. So my

0:49

first reaction was, ha ha, this is

0:51

so funny. And my next reaction was,

0:53

wait a minute, I'm a journalist, is

0:55

this real? And I think we will

0:58

see it to a streamer, President, maybe

1:00

within our lifetimes, maybe within

1:02

our lifetimes. It's the I'm

1:05

in terms of life. It's the Breakfast

1:07

Club. The world's most dangerous

1:09

morning show. Hey! Angela E. is

1:11

kind of like the big sister that

1:13

always picks in the boy. That's not

1:16

how it goes. That's not how anything

1:18

goes. Yeah, me's really like a... of

1:20

the best DJs ever. But leave that! Sean

1:22

Lamine is the wildcard. And I'm

1:25

about to give somebody the credit

1:27

they deserve from being stupid. I

1:29

know that's right. What's wrong with

1:31

you? Listen to the Breakfast Club

1:33

weekday mornings from 6 to 10 on

1:35

106 7 the beat. Columbus is real

1:37

hip popping on MB. Hi

1:40

everyone, before we get to the episode,

1:43

I just wanted to lead in and

1:45

say, we are up for a webby,

1:47

I'll be including a link, I know

1:50

it's a pain in the ass, to

1:52

register for something, I'm sorry, I really

1:54

want to win this, never want an

1:56

award in my life. It will be

1:59

in the links, and while you're there

2:01

and registered, look up the wonderful weird

2:03

little guys with Miss Molly Conger, vote

2:05

for both of us, I'm in the

2:08

Best Business podcast episode. One, she's in

2:10

the best crime podcast, episode one. We

2:12

can win this, we can defeat the

2:14

others. And now for the episode. Every

2:18

day I'm punished and killed and you

2:20

love to watch. Welcome to Better Offline.

2:22

We're live from New York City. Recorded

2:24

straight to tape, of course. And I'm

2:27

joined by an incredible cast of people.

2:29

To my right, I have Paris Martin,

2:31

I have the information aiding Paris. What's

2:34

up? What is up? Edward Anguoso of

2:36

the tech bubble newsletter. Hello, hello. And

2:38

the wonderful Allison Murrow of the CNN

2:40

nightcap newsletter. Hi. And Allison, you wrote

2:43

one of my favorite bits of my

2:45

favorite bits of media criticism I've ever

2:47

read criticism I've ever read recently. Do

2:49

you want to actually walk us through

2:52

that pace? Because I think I will

2:54

link it in the notes. Don't worry

2:56

everyone. I'd be happy to. I wrote

2:58

a piece. I think the headline we

3:01

ended up with was like, Apple's AI

3:03

is not the disappointment. AI is the

3:05

disappointment. Yeah. And this was inspired by

3:08

credit to where it's due. I was

3:10

listening to Hard Fork with Kevin Roos

3:12

on our, or my husband and I

3:14

were driving out to the country and

3:17

listening to this and just getting infuriated

3:19

infuriated. And basically their premise was, or

3:21

at least Kevin Ruz's premise, was that

3:23

AI is failing, or sorry, that Apple

3:26

is failing this moment in AI. And

3:28

Apple has been trying, it's been like

3:30

the laggard, you know, that's a narrative

3:32

we've heard in tech media over and

3:35

over. And it's like, Kevin Ruz's point

3:37

was like, oh, well, they should just

3:39

start getting more comfortable with experimenting with

3:42

experimenting and making mistakes and, you know.

3:44

Violating everything that Apple brand kind of

3:46

stands for and like force the AI

3:48

into a consumer product that no one

3:51

wants. And I was like respectfully, no.

3:53

It's just such a funny argument given

3:55

that it was a mistake being made

3:57

by Apple that resulted in the whole

4:00

hoofy PC small group situation. What was

4:02

there? Walk us through that. That was

4:04

specifically how the editor-in-chief of the Atlantic

4:06

ended up in a secret military signal

4:09

chat. Wait, I missed what? How have

4:11

you listened to this? I've met this

4:13

too. I'm sorry, have you guys not

4:16

been online? I don't use the computer.

4:18

Better online. Or better. Oh gosh, I should

4:20

leave them. I've been reading the scrolls.

4:23

So basically the Atlantic came out

4:25

a couple weeks ago with an

4:27

article about how their editor-in-chief one

4:29

day was suddenly added. to a

4:32

signal group chat. Right, signal game.

4:34

Yeah, signal game. But how did this

4:36

Apple lead to this? So the Apple

4:38

thing was, I'm forgetting who exactly reported

4:40

this, this was in the last couple

4:42

of days, that. how it happened was

4:44

the con, like you know that thing

4:46

that comes up in your iPhone

4:48

where it says like, oh a

4:50

new phone number has been found.

4:53

Yeah, it was a suggested contact

4:55

and it happened because someone I

4:57

guess in the government had copied

4:59

and pasted an email containing the

5:01

editor-in-chief of the Atlantic's contact information

5:03

in a message to, I'm forgetting

5:06

whatever government official. One of the

5:08

guys. Yeah, one of the guys.

5:10

And so he ended up combining.

5:12

the Atlantic EIC's information into a

5:14

contact for some government, dude. And

5:16

that's how they ended up in, because

5:19

it then signal when you connected to

5:21

your contacts. I love the computer so much.

5:23

So I mean. That makes me even crazier

5:25

about the hard fork take because it's like

5:27

you can't mess around with something like your

5:29

phone. Well in this particular instance I take

5:32

it all back Apple AI is amazing. It

5:34

gave us one of the best journalism stories

5:36

of the year. You also made a really

5:38

good point in here is message you this

5:40

on the way and you say you make

5:42

this point that there's a popular adage in

5:44

policy circles that the party can never fail

5:46

it can only be failed. It is meant

5:48

as a critique of the ideological gatekeepers who

5:50

may be, for example, blame voters for their

5:52

parties failing rather than the party itself. The

5:54

same fallacy is taking root among AI's biggest

5:56

backers. AI can never fail. It can only

5:58

be failed. And I love this because it's

6:00

you get people at Kevin Roos and

6:03

there was a wonderful clip on the

6:05

New York Times tic-tock of Kevin Roos

6:07

seeming genuinely pissy. He's like I can't

6:09

believe people are mad at AI because

6:11

of Siri and it's like oh what

6:13

they think it's shitty because it's shitty

6:15

like it's their child him and Casey

6:17

act as if we've hurt chat GP.

6:19

Sorry. Claude, they're anthropic boys. And, you're

6:21

saying that Casey's boyfriend works anthropic, I

6:23

know he does the site, it's fucking,

6:26

anyway. It's just so weird because it's

6:28

like, we have to apologize for not

6:30

liking AI enough, and now you have

6:32

the CEO of Shopify saying, actually, you

6:34

have to use it, do you hear

6:36

about this? Yeah, he said what, that

6:38

you have to prove your job can't

6:40

be replaced by AI, or else it

6:42

will be? And he also said that

6:44

now it's going to be Shopify policy

6:47

to include and all of the employee

6:49

performance reviews, both for your self-assessment and

6:51

for your like direct reports and colleagues

6:53

assessment, how much this person use AI?

6:55

And obviously, what's going on there is

6:57

if you are not reporting that you

6:59

use AI all the time for everything,

7:01

you could get fired question. I think

7:03

the line I just tried to overhaul

7:05

hiring practice so that they could have

7:08

AI first or AI only and then

7:10

roll it back because they realized you

7:12

can't replace you can't replace. these jobs.

7:14

My question is, I mean, this is

7:16

something that's brought up on the show

7:18

all the time, but who are these

7:20

people that are encountering the AI assistants

7:22

suddenly plugged into every app and being

7:24

like, yeah, this is actually beneficial to

7:26

my life and this works really well,

7:29

because it sucks every time I use

7:31

it. Or you made the point in

7:33

your article, Allison, as well, it's like,

7:35

if it was 100% accurate, it would

7:37

be really useful. If it's even 98%

7:39

accurate, it's not. I think that was

7:41

the point that, you know, to his

7:43

credit, Casey Newton made in the episode,

7:45

which is that AI is fundamentally an

7:47

academic project right now. And it's like,

7:49

yeah, we can have all the kinds

7:52

of debates about its utility, but ultimately,

7:54

is it a consumer product? And no,

7:56

it's just like it's. failing as a

7:58

consumer product on all fronts. And what's

8:00

crazy as well is, I'm surprised he

8:02

would say that considering everything else he's

8:04

ever said, because he quite literally has

8:06

had multiple articles recently being like, ah,

8:08

consumer adoption is up. He had an

8:10

article the other day where it was

8:13

like, Data provided exclusively from anthropic shows

8:15

that more people are using I am.

8:17

It's like my man. It's 2013 again

8:19

We're past this. You can't just do

8:21

this anymore and lest you or you

8:23

and so going back to mr. Mr.

8:25

Lootkey of Shopify I just want to

8:27

read my favorite part of it. It

8:29

says I use it all the time,

8:31

but even I feel I'm only scratching

8:34

the surface, dot, dot, dot. You've heard

8:36

me talk about A.I.M. Weekly videos, podcast,

8:38

Townhalls and Summit. Last summer, I used

8:40

agents to create my talk and presented

8:42

about that. So all this fucking piss

8:44

and vinegar and the only thing you

8:46

can use it for is to write

8:48

a slop-ridden presentation to everyone about how

8:50

good AI is without specifying what it

8:52

does, I feel like I'm going insane

8:55

sometimes with this stuff. I mean, in

8:57

one way that's great, right? The only

8:59

place you should encounter it is maybe

9:01

the team building retreats, you know, that's

9:03

the utility of this shit. This reminds

9:05

me a lot of like media 2012,

9:07

2013 where it was all pivot to

9:09

video and what's our vertical video strategy

9:11

and it's strategy and it's like... Okay,

9:13

now what's our AI strategy? How are

9:15

we injecting AI into everything we're doing?

9:18

And it's like, well, to what end?

9:20

What's the point? This is something that

9:22

has been driving me. Mad, especially with

9:24

partnerships we're seeing between media firms and

9:26

these AI firms. You know, these are

9:28

firms in the same sector that keeps

9:30

lying to firms about how if you

9:32

integrate artificial intelligence, this time it'll optimize

9:34

your ability to find an audience or

9:36

to get revenue and we can include

9:39

you in some esoteric revenue share program

9:41

or we'll be able to, you know,

9:43

claw back some of the eyeballs and

9:45

the attention that you're interested in seeking.

9:47

But... Each time it's actually just used

9:49

to graph themselves on them. or to

9:51

try to gin up excitement about these

9:53

products, right? What's insane is this company

9:55

has a multi-billion dollar market cap. And

9:57

I'm just going to read point two.

10:00

AI must be part of your GSD

10:02

prototype phase. The prototype phase of any

10:04

GSD product should be dominated by AI

10:06

exploration. Prototypes are meant for learning and

10:08

creating information. AI dramatically accelerates this process.

10:10

How? Fucking how! Like that's the thing.

10:12

I have plans on my PR phone

10:14

but will occasionally bring the AI things

10:16

and I'm every time I'm just like

10:18

this better fucking work. Like just every,

10:21

and to the credit they do, but

10:23

it's like I have clients I turn

10:25

down all the time and I'm like,

10:27

yeah we're doing this. And I'm like,

10:29

is this just the chat box? And

10:31

they're like, no, I'm like, can you

10:33

show me how it works? No. I'm

10:35

like, oh cool, yeah, don't think we're

10:37

going to be a good fit somehow,

10:39

because you don't seem to be able

10:41

to explain what your product does. But

10:44

don't worry, this appears to be a

10:46

problem up to the multi-billion dollar companies

10:48

as well. It's just, it feels like

10:50

the largest mask-off dunce moment in history.

10:52

Just these people who don't do any

10:54

real work, being like, I don't do

10:56

anything real. And the pivot-to-video I think

10:58

is actually a really good comparison, I

11:00

don't think I want to consume video

11:02

in the way that what it was

11:05

like Mike and everyone and they were

11:07

like oh we're going to do this

11:09

video and this we're going to do

11:11

everything video now video first no written

11:13

content it's like I don't know a

11:15

single god damn human that actually does

11:17

that and also the other thing that

11:19

Facebook was just over just claiming like

11:21

averaging out the engagement numbers and everyone

11:23

was wrong. But that was the same

11:26

kind of thing. It's like very clearly

11:28

the people who have their hands and

11:30

the steering wheel are looking at their

11:32

phone. And it's fucking confusing, but it's

11:34

so much worse this time. It feels

11:36

more egregious somehow. Yeah, because it feels,

11:38

I mean, we've had so many of

11:40

these hype cycles kind of back to

11:42

back to back to even the horizontal

11:44

video days of vertical video, to whatever

11:47

the hell the universe was supposed to,

11:49

literally in a... ominous moment

11:51

as I was

11:53

walking in to record

11:55

this, I saw

11:57

a guy wearing a

11:59

leather jacket with

12:01

Bordet Biot Club on

12:03

the back and

12:05

I was like, God,

12:07

what a cool dude,

12:12

but it's like, how long is this going

12:14

to last? I

12:16

have been actually looking at the

12:18

numbers recently and I don't know

12:20

either because for SoftBank to fund

12:23

OpenAI might require them to

12:25

destroy SoftBank, like S &P

12:27

is downgrading the credit rating potentially due Hell

12:29

yeah. Yeah, I know, we're really at this

12:31

point where it's just like, we've gone so

12:33

much further than like the metaverse and crypto

12:35

do because those weren't really like systemic things,

12:37

but this one, I think it's just the

12:39

narrative has carried away so far that people

12:41

are talking about a thing that doesn't exist

12:43

all the time. I mean

12:45

in some elements, it kind

12:48

of reminds me at the near

12:50

the end or near the

12:52

real peak is when we started

12:54

also to see metaverse and

12:56

crypto sustainable refi shit where actually

12:59

we can fight climate change

13:01

with crypto, but putting carbon credits

13:03

on the blockchain and so

13:05

there was a moment where the

13:07

frenzy and the speculative frenzy

13:09

led to like world transformative visions

13:12

that were bullshit and I

13:14

feel like we are heading

13:16

there, we're in that

13:18

direction with artificial intelligence

13:20

where consistently we've been fed, oh,

13:22

this is going to revolutionize everything, but

13:24

it feels like the attempt to graft

13:27

it onto more and more

13:29

consumer products, more and more government

13:31

services, more and more parts

13:33

of daily spheres of life as

13:35

a way to like privatize

13:38

almost everything or commodify everything feels

13:40

like downstream of the way

13:42

cryptos attempt to put everything on

13:44

the blockchain blew up. Yeah,

13:46

I was thinking about this in

13:48

a kind of like fundamentally cultural way

13:50

where I think at some point

13:53

in the last 30 years, there was

13:55

a time when everything coming out

13:57

of Silicon Valley was cool, whether it

13:59

was like or world transformative, it was

14:01

cool and there was like an

14:03

edge to it. And people were

14:06

like, ooh, that's neat. Disruptive.

14:08

Yeah, disruption was everything.

14:10

And like, I think post like

14:12

Facebook Cambridge Analytica era,

14:14

like 2016, Tech has just

14:16

stopped being cool and edgy. It's

14:19

very corporate and like. I don't think

14:21

the rest of corporate America has kind

14:23

of figured out that Silicon Valley is

14:25

not the cool thing anymore. And that

14:27

they're fully capable of being wrong and

14:29

lying. Like that's the other thing. They've

14:32

gotten very good at fundraising and

14:34

marketing. But they're also not like kids

14:36

anymore. Like we talk, I still see people

14:38

referring to open AI as a startup. Palmer

14:41

Lucky is a kid. Palmer Lucky

14:43

is a kid who looks like

14:45

leisurely, Larry, and sells arms. Which

14:47

is a US government. That will

14:49

be A Howard. Just a little

14:51

guy. Well, we refer to them

14:53

as startups, but also, I think one

14:56

of the most accomplished parts of

14:58

AI marketing has been like we.

15:00

always refer to them as labs.

15:02

Yeah. So they seem like so

15:04

academic and like good fundamentally and

15:06

it's like these are companies like

15:08

some of them might be part

15:10

of you know a research institution or

15:13

a university but a lot of

15:15

them are startups. Yeah literal

15:17

companies. Yeah they are companies

15:19

like anthropics public benefit I

15:21

believe and it's just remarkable

15:23

and I think what's happened here

15:25

is that the narrative has gotten

15:28

away to the point that We're really, the

15:30

dance mask off moment I mention

15:32

is people like Mr. Luque from

15:34

Shopify. It's very clear he

15:36

doesn't do any work. Like I think

15:39

that anyone who is just being

15:41

like, yeah, A.I. is the future

15:43

and it's changing everything without specifying

15:45

anything, doesn't do any work. I

15:47

just don't, Bob I. Go from

15:49

Disney said A. I was going

15:51

to check, no, it's not, Bob,

15:53

how is it changing your fucking life,

15:55

you, you lazy past, like, like, like,

15:58

and... It's just, it's so bizarre. But

16:00

it feels like we're approaching this

16:02

insanity level, where you've got people

16:04

like, Shopify being like, oh yeah, it's going

16:06

to be in everything, as like open AI

16:08

burns more money than anyone's ever burned,

16:10

Anthropic lost 5.6 billion last year

16:13

report by the information. It does

16:15

some incredible fucking work on this, I

16:17

should say. And it just doesn't make any

16:19

sense. And it's getting more nonsensical. You're seeing

16:21

like all of the crypto guys have fully

16:23

become AI guys now, and that was something

16:25

I didn't like talking about at first, because

16:27

it wasn't happening in the... now it's all

16:30

of them. They all have AI avatars. This

16:32

guy called Jamie Burke is real, real shithead.

16:34

This guy was like a crypto-metiverse guy, and

16:36

he's now a full AI guy. Another guy

16:38

called Bernard Maher, who is just a harmless

16:40

Forbes. like a kind of like an NPC

16:43

like one of the hollows from Dark Souls

16:45

walking around diagram is increasingly

16:47

becoming a circle yeah but he's on

16:49

to quantum now which is a bad sign

16:51

that's a bearish sign when you got one

16:54

of the Forbes guys moving on to quantum

16:56

we cooked What about thermal? Isn't it?

16:58

Who's thermal? Isn't there some like, uh...

17:00

Oh, thermal, yeah. There's some scam. Thermodynamics,

17:02

fuck a unit because of thermodynamics influence.

17:04

So I know what that means. I

17:06

also know what that means, but if

17:09

anyone could tell me real quick. But

17:11

it's, I think the most egregious one

17:13

I've seen, I sent this all to

17:15

you. And I think you and I

17:17

have talked about this the most. There

17:19

was one of the stupidest fucking things

17:21

I've read in my worthless life. And

17:23

it's called AI 2027. Now, if you

17:26

have not run into this yet as

17:28

a listener, it will be in the

17:30

episode notes. I'm just going to bring

17:32

it up because it is fan fiction.

17:34

It is literally big. Throughout I was

17:37

like, is this fan fiction? This is

17:39

fan fiction. Oh. This is interactive fan

17:41

fiction. Oh. This is interactive fan fiction.

17:43

will future AI agents have can be

17:46

found here. The scenario itself is written

17:48

iteratively. We wrote the first period up

17:50

to mid- 2025, then the following period,

17:52

etc. until we reach the ending. Yeah,

17:55

otherwise known as how you write stuff. Like you write

17:57

it writing an ill and near fashion. We then

17:59

scrap this. and did it again, you

18:01

should have scrapped it in all of

18:03

it. Now this thing is, it's predicting

18:05

that the impact of superhuman AI over

18:08

the next decade will be enormous exceeding

18:10

that of the Industrial Revolution. We wrote

18:12

a scenario that represents our best guess

18:14

that what it might look like otherwise

18:16

known as making stuff up. Not even

18:19

over the next decade, it basically says

18:21

it's going to have superhuman like... catastrophic

18:23

or world-changing impact from the next five

18:25

years. Like by 2023, we're either going

18:27

to be completely overtaken by our robot

18:30

overlords or like at a tenuous piece.

18:32

Yeah. And it's insane as well because

18:34

it has some great headlines like mid-2020,

18:36

China wakes up. I love that China

18:38

was so far behind. You know, it's

18:41

much like... When did this come out?

18:43

This came out like a week ago

18:45

and I've been sent it by a

18:47

lot of people. If you're one of

18:49

the people who sent it, don't worry,

18:52

I'm not mad at you. It's just

18:54

I got sent it by a lot

18:56

of people. This thing is one of

18:58

the most well-written pieces of fan fiction

19:00

ever in that it appears to be

19:03

like a Mancharian candidate situation for idiots.

19:05

Not saying this is about the same

19:07

thing. He wrote up a piece about

19:09

this called the AI forecast predicts some

19:11

storms ahead. Some storms. Some storms. That's

19:13

not even an accurate description of all

19:16

of the storms it predicts. And the

19:18

long and short of this, by the

19:20

way, I have read this a few

19:22

times because I fucking hate myself. The

19:24

long and short of it is that

19:27

a company called Open Brain, who could

19:29

that be? Yeah. Could be anyone. Anyone.

19:31

Open brain. They create a self-learning agent

19:33

somehow. Unclear how. All they add is

19:35

just that how many are terror flops

19:38

it's going to require. And it can

19:40

train itself and also requires more data

19:42

centers than ever. How they get them,

19:44

how those are funded, no fucking clue

19:46

isn't explained. Probably the easy, actually, this

19:49

just occurred to me. Probably the only

19:51

thing. could actually reasonably extrapolate in here

19:53

is the cost of data centers. That's

19:55

the only thing and they don't. Probably

19:57

because they'd be like yeah we need

20:00

a trill an actual trillion dollars to

20:02

do this made-up thing. I do also

20:04

want to add in here that you

20:06

know behind the AI 2027 is you

20:08

know one of the people connected to

20:11

it if I remember correctly is Scott

20:13

Alexander who's this Guy, that's part of

20:15

the rationalist community, which is one of

20:17

the, uh, what are the groups that

20:19

overlaps with effective altruists, the accelerationist. Yeah,

20:22

you know, so if it feels like

20:24

it's, uh, frothy and fan fictionee and

20:26

hypi, that's because these are the same

20:28

people that keep. that are connected to

20:30

pushing constant hype cycles over and over

20:32

and over again. And it's written to

20:35

be worrying as well. It's written to

20:37

up save. It's written to be worrying,

20:39

but it also in the predictions for

20:41

the next two years keeps talking about

20:43

how the stock market is going to

20:46

grow exponentially and do so well. The

20:48

president is going to be making all

20:50

of these wise informed decisions and having

20:52

really deep conversations with the leader of

20:54

open brain. And I was like, are

20:57

you, that's why I asked when did

20:59

this come out. of years ago. But

21:01

no, it is why, like literally it

21:03

says... It's like the men of Jesuits

21:05

kind of planning like a quiz. That's

21:08

out out of it, he's gonna be

21:10

born, he's gonna lead us to the

21:12

promise list. It's so good as well,

21:14

because the people who've sent this to

21:16

me have been very concerned, just because

21:19

they're like, this sounds scary. And I

21:21

really want to be clear, if you

21:23

read something like this and you're like,

21:25

that doesn't make sense to me. The

21:27

AI R&D progress multiplier, what do we

21:30

mean by 50% faster algorithmic progress? We

21:32

mean that open brain makes as much

21:34

AI research progress in one week with

21:36

AI as they would in 1.5 weeks

21:38

without. Who fucking cares, Matt? What are

21:40

you talking about? If a frog had

21:43

wings, he could fly. Like, what you...

21:45

And what's crazy is, and I know

21:47

I'm back on Kevin Roos, it's because

21:49

he's a nakedly captured part of the

21:51

tech industry now. I am in public

21:54

relations and I'm somehow less frothy about

21:56

this, that you tell you fucking everything.

21:58

It is insane that the New York

22:00

Times, at a time when you have

22:02

soft bank being potentially downgraded by S&P,

22:05

you have open AI raising more money

22:07

than they're ever raised, 40 billion dollars,

22:09

except they only receive 10 billion, and

22:11

they'll only get 20 billion more by

22:13

the end of the year if they

22:16

become a non-for profit, which they can't

22:18

do. No, no, no, no, Kevin can't

22:20

possibly cover that. I'll swipe, I can't

22:22

even get my phone to it. I

22:24

do really love all of the incredibly

22:27

solid already feature profiles. I'll put the

22:29

link in there for this. He's got

22:31

75 pockets on his tour. Yeah, my

22:33

man is ready to... Oh, that's where

22:35

it is. And it's just him, like,

22:38

this guy's sitting with his hands class,

22:40

like, staring mournfully into the distance. This

22:42

is what you're spending your time on,

22:44

Kate. And I'm just going to read

22:46

some Kevin Rose. The AI-I prediction well

22:49

is torn between optimism and gloom. A

22:51

report released on Thursday decided lands on

22:53

the side of gloom. That's Ken Roos'

22:55

voice. But my favourite part of this,

22:57

by far. I'm going to take a

22:59

second to get it because Ed I

23:02

sent this to you as well. Where

23:04

is it? So also a lot of

23:06

this is... Oh, here we go. If

23:08

all of this sounds fantastical, well it

23:10

is! Nothing remotely like, who want Mr.

23:13

Kokker to tell Joe and Mr. Liffland

23:15

of predicting as possible with today's AI

23:17

tools, which can barely order a burrito

23:19

and dordash without getting stuck. Thank you

23:21

Kevin. I'm so fucking glad the New

23:24

York Times is on this. And that

23:26

was at the end of the, like,

23:28

the altruistic AI guys. all have told

23:30

themselves this story and they all believe

23:32

it and they think they are like

23:35

the Prometheus bringing fire to the people

23:37

and like warning the people and it's

23:39

like you guys have sold yourself a

23:41

story with no proof. I don't know

23:43

I feel like they just scam artists

23:46

that nothing about this suggests they believe

23:48

in anything. You can just say stuff.

23:50

Look, it works. Literally. The second sentence

23:52

in this is that in two months,

23:54

there will be personal assistance that you

23:57

can prompt them with tasks like order

23:59

me a burrito on Dordash and they'll

24:01

do great stuff. There are so. So

24:03

many things that go into ordering me

24:05

a burrito and door dash. What restaurant

24:08

do I want? What burrito do I

24:10

want? How do I want it to

24:12

get to me? Where am I? It

24:14

can't do any of those things, nor

24:16

will it. He gazed out the window

24:18

and admitted he wasn't sure, and the

24:21

next few years went sure, and the

24:23

next few years went well and we

24:25

kept well and we kept aye out

24:27

of the window and admitted he wasn't

24:29

sure, and the next few years went

24:32

well and we kept well, and the

24:34

next, and the next few years went

24:36

well, and the next few years went

24:38

well, and the next few years went

24:40

well, and... something like that. You know

24:43

one of the things I really really

24:45

love about I don't know it's just

24:47

it's so frustrating because we're we're constantly

24:49

fed these you know sci-fi esoteric futures

24:51

about how AI, powerful AI, superhuman AI,

24:54

is around the corner, and we need

24:56

to figure out a way to accommodate

24:58

these sorts of futures. We need, and

25:00

part of that accommodation means restructuring the

25:02

regulations we have around it, part of

25:05

that accommodation means entertaining experiments, grafting them

25:07

onto our cultural production, grafting them onto

25:09

consumer goods, part of that means just

25:11

like, you know, you know, taking it

25:13

on the chin and figuring out how

25:16

to use statue BT. But in all

25:18

of this. Just more or less sounds

25:20

like you need to, the marketing is

25:22

failing on you and you need to

25:24

step up. Yes, you need to believe.

25:27

You need to believe in this. You

25:29

need to do your part, you know,

25:31

to summon God. And that's the thing.

25:33

It goes back to what you're saying.

25:35

It's like you've failed AI by not

25:37

believing. Yeah, and if you're bad at

25:40

it, it's your fault and not the

25:42

machine's fault. And to learn. To Ed's

25:44

point, I think like all of this

25:46

like, like, predicting of the future, like,

25:48

like, like, like, They have told themselves

25:51

a story that is, this is inevitable.

25:53

And that there are no choices that

25:55

the human beings in the room get

25:57

to make about how this happens. And

25:59

it's like, actually, no, we can make

26:02

choices about how we want our future

26:04

to play out. And it's not going

26:06

to be just Silicon Valley shoving it

26:08

down our throat. And on the subject

26:10

of human choice, if this shit is

26:13

so powerful, why have their mighty human

26:15

choices not made it useful yet? Like

26:17

that's the thing. It's... And you make

26:19

this point in your piece as well.

26:21

It's like, hey I can ever fail,

26:24

can I be failed? Failed by you

26:26

and me, the smooth-brained luddites who just

26:28

don't get it. And it's like, why

26:30

do I have to prove myself? And

26:32

listen, you know, the luddites, they have

26:35

more grooves on their brain than Kevin.

26:37

So, I think it's worth embracing a

26:39

little bit, you know? Ryan

26:48

Reynolds here for Mint Mobile.

26:50

The message for everyone paying

26:52

big wireless way too much.

26:54

Please for the love of

26:56

everything good in this world,

26:58

stop. With Mint you can

27:00

get premium wireless for just

27:03

$15 a month. Of course if

27:05

you enjoy overpaying, no judgments,

27:07

but that's joy overpaying.

27:09

No judgments, but that's

27:11

weird. Okay, one judgment.

27:13

Anyway, give it a

27:15

try at mintmobile.com, Hi,

27:19

I'm Morgan Sun, host of Close All

27:21

Tabs from KQEDD, where every week we

27:24

reveal how the online world collides with

27:26

everyday life. You don't know what's true

27:28

or not, because you don't know if

27:30

AI was involved in it. So my

27:33

first reaction was, ha-ha, this is so

27:35

funny, and my next reaction was, wait

27:37

a minute, I'm a journalist, is this

27:39

real? And I think we will see

27:42

it to a streamer, President, maybe within

27:44

our lifetimes, maybe within our lifetimes. Hey

27:47

Zucco and Kayla from the wake-up call enjoy

27:49

your podcast when you're done. Don't forget about

27:51

us We have a radio show. We try

27:53

to bring a smile to your face every

27:56

morning We also talk to some of the

27:58

hottest country stars of today and we like

28:00

to share some good news with that's what

28:02

I like. Because Lord knows that's hard to

28:05

find. When you're done podcasting your podcast listen

28:07

to us at 92.3 W.C.O.L. Set your preset

28:09

on your radio right now and don't forget

28:11

you can listen to us online on

28:13

the iHAR radio app. And look yeah

28:15

I feel like Rob Horning wrote this

28:17

newsletter a few weeks ago that I

28:19

think he was honing in on this

28:21

point that LLLams and these generative

28:23

AI chat bots and the tools that

28:26

come out of them are in some

28:28

ways a distraction because a lot of

28:30

these firms are pivoting towards how do

28:32

we you know create all these products

28:34

but also how do we figure

28:36

out you know government products that

28:38

we can provide right how do

28:40

we get into defense contracting how

28:42

do we get into arming or

28:44

integrating AI into arms and and

28:47

increasingly it feels like You know,

28:49

yeah, your AI agent's going to

28:51

be able to, not going to

28:53

be able to order your burrito.

28:55

But these firms are also, you

28:57

know, at the same time that

28:59

they're insisting superhuman intelligence is around

29:01

the corner and we're going to

29:03

be able to make your individual

29:06

lives better, are spending a lot

29:08

of time and energy on use

29:10

cases that are actually dangerous, right?

29:12

And it should actually be concerning,

29:14

but they, but... the firms that

29:17

are offering these generative products are

29:19

spending actual you know the stuff

29:21

that they're actually putting their time

29:23

and energy into is you know the

29:25

sort of demonstrably destructive tools under the

29:27

guys in the in the kind of

29:29

murky covering of it's all you know

29:32

artificial intelligence right it's all inevitable it's

29:34

all coming down the same pipeline you

29:36

should accept it yeah and it's I think

29:38

the thing is as well is those guys really

29:40

think that's the next big money maker, but I

29:42

don't think anyone's making any money off of this

29:45

No one wants to talk about the money because

29:47

they're not making any Like no one like I

29:49

think I've read the I've read the earning schools

29:51

and not going to listen of every single company

29:53

that is selling an AI service at this point

29:55

I can't find a single one that wants to

29:57

commit to a number of the Microsoft and they'll

30:00

only talk annualized, which is my favorite

30:02

one. ARR. AR. AR. AR. AR. AR.

30:04

AR. But the thing is, ARR traditionally

30:06

would mean an aggregate rather than just

30:09

12 times the last biggest month, which

30:11

is what they're doing. No, that's the

30:13

classic setup. Used to be an ARR.

30:16

No, I refuse to. No, my client's

30:18

ass is to the ground with that

30:20

one, because it's like you can't just

30:23

fucking make up a number, unless you're

30:25

an AI than you absolutely can. other

30:27

than all the others I've listed, is

30:29

that I feel like in their position

30:32

and in the position of anyone with

30:34

any major voice in the media, skepticism

30:36

isn't like something you should sometimes bring

30:39

in. It's like you don't have to

30:41

be a grizzled hater like myself, but

30:43

you can be like, hey, even if

30:46

this did work, which it doesn't, how

30:48

does this possibly last another year? And

30:50

the reaction is, no, actually it's perfect

30:52

now and will only be more perfect

30:55

in the future. And I still get

30:57

emails from people, because I said once

30:59

on an episode, if you have a

31:02

use for AI, please email me. Regret

31:04

of mine. Every time I get an

31:06

email like this, so it's very simple.

31:09

I've set up seven or eight hours

31:11

worth of work to make one prompt

31:13

work, and sometimes I get something really

31:15

useful. It saves me like 10 minutes.

31:18

And you're like, great. And what for?

31:20

It's like, oh, just some productivity things,

31:22

what productivity things, they stop responding. And

31:25

it's just. I really am shocked we

31:27

got this far. I'm going to be

31:29

honest. At this point, I will never

31:32

be tired because my soul burns forever,

31:34

but it's exhausting watching this happen and

31:36

watching how it's getting crazier. I thought

31:38

like as things got worse, people would

31:41

be like, well, CNN stepping up, but

31:43

it's like watching the times and some

31:45

parts of the journal still feed this.

31:48

Also, the journal has some incredible critical

31:50

work on that. It's so bizarre. The

31:52

whole thing is just so bizarre. and

31:54

has been so bizarre to watch an

31:57

attack media. I mean I think part

31:59

of it is also just because investors

32:01

have poured a lot of money into

32:04

this and so of course they're... going

32:06

to want to back what they have

32:08

spent hundreds of millions or billions of

32:11

dollars on. And much of the tech

32:13

media involves reporting on what those investors

32:15

are doing, thinking, and saying. And whether

32:17

or not what those people are saying

32:20

or doing, it's often not based in

32:22

reality. Yeah, I say as not a

32:24

member of the tech media. So I

32:27

have like kind of a general assignment,

32:29

business markets, Econ. That's kind of my

32:31

jam. When I come, when like, A.I.

32:34

first started becoming the buzzword, like, ChatGPT

32:36

had just come out, I was like,

32:38

oh, this sounds interesting. So I was

32:40

paying attention, like a lot of journalists

32:43

were. And, you know, like we've hit

32:45

limitations. And I think part of the

32:47

reason it's gotten so far is because

32:50

the narrative is so compelling, curing cancer.

32:52

Yeah. We're going to, we're going to

32:54

end hunger. Nice. Okay, how? How? Also,

32:57

the problem of hunger in the world

32:59

is not that we don't grow enough

33:01

food. It is a distribution problem. It

33:03

is a sociologic, it is a complicated

33:06

problem. What actually is AI going to

33:08

do? Also that you're going to need

33:10

human beings to distribute it just like

33:13

if you push them one step if

33:15

you read the 2027 AI thing It

33:17

explains that the AI is going to

33:20

give the government officials such good advice

33:22

They'll be actually really nice and caring

33:24

to work and what's crazy is here's

33:26

the thing and I'm glad you brought

33:29

up one thing I've learned about politics

33:31

particularly recently But in historic means when

33:33

the government gets good advice they take

33:36

it every time every time they're like

33:38

this will this is economically good like

33:40

Medicare for all which we've of course

33:43

had forever and never and came close

33:45

to numerous times decades ago versus now

33:47

when we have him and I think

33:49

the other funny thing is as well

33:52

with what you were saying Allison is

33:54

like yeah It's going to cure cancer.

33:56

Okay, can it do that? No. Okay,

33:59

it's going to cure hunger. Can it

34:01

do that? No. Okay, easy then. Perhaps

34:03

it could make me an appointment. Also,

34:06

no. Can you buy something with it?

34:08

No. Can it take this spreadsheet and

34:10

move stuff around? Maybe... Sometimes? It can

34:12

write a robotic sounding script for you

34:15

to make the appointment yourself. Wow. You

34:17

know. I mean, I would even say

34:19

that like, I could give... the benefit

34:22

of the doubt to researchers who are

34:24

really working on the scientific aspects of

34:26

this like I don't I'm not a

34:28

scientist I don't know how to cure

34:31

cancer but if you're working with an

34:33

AI model that can do it like

34:35

God bless but businesses actually do take

34:38

money-making advice and money-making technology when it's

34:40

available and I think about this all

34:42

the time with crypto which is another

34:45

area I cover a lot it's like

34:47

If it were the miracle technology that

34:49

everyone or its proponents have said it

34:51

is, businesses would not hesitate to rip

34:54

up their infrastructure to make more money.

34:56

And like no one's doing it. And

34:58

it's like, oh, well, they just haven't,

35:01

they haven't figured out how to optimize

35:03

it yet. And it's like, that sounds

35:05

like a failure of the product and

35:08

not a failure of people using it.

35:10

So I get back to the whole

35:12

like. Yeah, it cannot fail. It can

35:14

only be failed. And it's the same

35:17

with crypto and a lot of other

35:19

tech, where it's just like, this is

35:21

not a product that people are. are

35:24

hankering for. And I think part of

35:26

the notable thing is when we do

35:28

see examples of large businesses being like,

35:31

oh yeah, we're gonna change everything about

35:33

our business and integrate AI, we're gonna

35:35

be an AI first company, the products

35:37

that end up coming out of that

35:40

are there's an AI chat bot in

35:42

my one medical app now. Cool, that

35:44

does nothing for me. When I'm trying

35:47

to search the Amazon comments on a

35:49

product, suddenly the search box is replaced

35:51

with an AI chat bot. That's not.

35:54

doing even one-tenth of what you've promised.

35:56

It's just the same product every fucking

35:58

time. It's just an AI. chatbot that

36:00

isn't super helpful. And it's great. I

36:03

remember back in 2015, 2016, I had

36:05

an AI chatbot company. They took large

36:07

repositories of data and turned it into

36:09

a chatbot, you'd use. I remember pitching

36:11

reporters at the time, and then being

36:13

like, who fucking cares? Who gives a shit?

36:15

This will never be... Like, a decade

36:18

later, everyone's like, this is literally God.

36:20

I cannot wait to go to the office

36:22

of a guy who wrote fan fiction about

36:24

this and talk to him about how scared

36:26

I am now. I can't wait for AGI.

36:28

And I've also said this before, but what

36:31

if we make AGI, none of them are

36:33

going to do it doesn't exist, and it

36:35

didn't want to do any work? That's the

36:37

other thing, like they're not, they don't, Casey

36:39

Cagaw, a friend of the show made this

36:42

point, it's made it to me numerous times,

36:44

which is, they talk about AGI and Ruz

36:46

did this as well. Like, AGI, this, AGI,

36:48

that. They don't want to define it because

36:50

if you have to start defining AGI,

36:53

you have to start defining

36:55

AGI, you have to start

36:57

talking about things like person

36:59

legit. And hey, how many is one? Is

37:01

it one unit? Is it a virtual machine?

37:03

Like there are real tangible things. And you

37:05

know they don't want to talk about that

37:08

shit because you even start answering one of

37:10

those and you go, oh right, we're not

37:12

even slightly closer. We don't even know how

37:14

the fuck to do it. A single one

37:17

of these things ever. And I honestly, the

37:19

person I feel bad for this is a

37:21

joke is Blakele Mine. I think his name

37:23

was from Google. If he'd have come out

37:26

like three years later and said that he

37:28

thought the computer, this guy from Google who

37:30

thought the man, the AI there was... The

37:32

guy who was like, the chat butt

37:34

is real and I love it.

37:37

Yeah, had that come out three

37:39

years later, he'd be called Kevin

37:41

Roos, because that's exactly what Kevin

37:43

Roos wrote about being AI, it's

37:45

like being AI, told me to

37:47

leave my wife. And Kevin, if

37:49

you ever fucking here, this man,

37:51

you're worried about me do you, I'm

37:53

a... Gagit Gizmo God. I love my dude dad's

37:55

I love my shit. I really do if this

37:57

was gonna do something fun I'd have done it

37:59

like I I've really spent time trying and

38:01

I've talked to people like Simon Wilson,

38:04

Max Wolfe, two people who are big

38:06

L&M heads who are, who disagree with

38:08

me on numerous things, but their reaction

38:10

is kind of, I'm not going to

38:13

speak exactly for them, it is basically,

38:15

it actually does this, you should look

38:17

at this thing, it does not, this

38:19

is literally God, but it all just

38:21

feels unsustainable economically, but also I feel

38:24

like the media is in danger when

38:26

this halls a part too, because... The

38:28

regular people I talked to about chat

38:30

GPT, I pretty much hear two use

38:33

cases. One, Google Search isn't working, and

38:35

two, I need someone to talk to,

38:37

which is a worrying thing. But, and

38:39

I think by the way, that use

38:42

case is just, that's a societal thing,

38:44

that's a lack of community, lack of

38:46

friendship, lack of access to mental health

38:48

services, and also could lead to some

38:51

terrible outcomes. But for the most part.

38:53

I don't know why I said for

38:55

the most part. I've yet to meet

38:57

someone who uses this every day and

39:00

I've yet to meet someone who really

39:02

cares about it. Like if I didn't

39:04

have my little anchor battery packs, I'd

39:06

scream. If I couldn't have permanent power

39:08

everywhere. Like if I couldn't have permanent

39:11

power everywhere. Like if I couldn't like

39:13

listen to musical day, that'd make me

39:15

real sad. If I couldn't access chat,

39:17

GPT, I don't know. I feel like

39:20

people's response to the media is going

39:22

to be negative too because there's so

39:24

many people that boosted it. There was

39:26

a verge story. There was a study

39:29

that came out today, I'll link it

39:31

as well in the notes, where it

39:33

was a study found that most people

39:35

do not trust, like regular people do

39:38

not trust AI, but they also don't

39:40

trust the people that run it, and

39:42

they don't like it. And I feel

39:44

like this is a thing that the

39:47

media is going to face at some

39:49

point. And Roose this time. Baby you

39:51

got away with the crypto thing you're

39:53

not this time I'm going to be

39:55

hitting you with the TV off to

39:58

share every day. But it's just I

40:00

don't think members of the media realize

40:02

the backlash is coming and when it

40:04

comes it's going to truly it is

40:07

going to lead to an era of

40:09

cynicism, true cynicism in society that's already

40:11

growing about tech, but specifically I think

40:13

it will be a negative backlash to

40:16

the tech media. And now would be

40:18

the great time to unwind this, versus

40:20

tripling down on the fan fiction that,

40:22

and I have been meaning to read

40:25

this out, my favorite part of this,

40:27

by far, I say it, and of

40:29

course, flawlessly have this ready, why our

40:31

uncertainty increases substantially beyond 2026? Our forecast

40:33

from the current day through 2026 is

40:36

substantially more grounded than what follows. Thanks

40:38

Motherfasm. Awesome! That's partially because it's nearer.

40:40

But it's also because the effects of

40:42

AI on the world really start to

40:45

compound in 2027. What do you mean?

40:47

They don't. You're claiming that. And I

40:49

just, I also think that there's the

40:51

greatest subtle problem that we have too

40:54

many people who believe the last smart

40:56

person they listened to and I say

40:58

that as a podcast runner. Like the

41:00

last invest that they talked to, the

41:03

last expert they talked to someone from

41:05

a lab. Yes, yes. Well, I think

41:07

that gets to, if you just push

41:09

the proponents, and this is like, I've

41:12

come into AI skepticism as a true,

41:14

like, I'm interested in this, I'm interested

41:16

in what you're pitching to the world,

41:18

and when I hear, and I hear

41:20

like, CEOs of AI firms get interviewed

41:23

about this all the time and they

41:25

talk about this future where everyone just

41:27

has a life of leisure and we're

41:29

lying around writing poetry and touching grass

41:32

and like everything's great no one has

41:34

to do hard labor anymore they have

41:36

that vision or they have like the

41:38

you know P doom of 75 and

41:41

everything is going to be terrible and

41:43

but no one has a really good

41:45

concept and that's why this is so

41:47

funny the fan fiction of like what

41:50

happens in 2027 it's like no one

41:52

has laid out any sense of like

41:54

how the job creation or destruction will

41:56

happen like in this piece they say

41:59

like oh there's gonna be more jobs

42:01

in different areas but some jobs have

42:03

been lost and it's like how why

42:05

they get what job They get oddly

42:07

specific on some things, then the meaningful

42:10

things, they're like, yep, there'll be jobs.

42:12

Yeah, and the stock market's just going

42:14

to go up. And the number go

42:16

up all the time, as it is

42:19

right now, it's really important. Yeah, I

42:21

believe they say in 2028. Agent 5,

42:23

which is the super AI, is deployed

42:25

to the public and begins to transform

42:28

the economy. People are losing their jobs,

42:30

but Agent 5 instances in the government

42:32

are managing the economic transition so adroitly

42:34

that people are happy to be replaced.

42:37

GDP growth is stratospheric. Government tax revenues

42:39

are growing equally quickly, and Agent 5

42:41

advised positions show an uncharacteristic generosity towards

42:43

the economically dispossessed. You know what this

42:46

is? We failed to uphold like public

42:48

arts education in America and a bunch

42:50

of kids got into coding and know

42:52

nothing but computers and so they can't

42:54

write fan fiction. Yeah. No one's fucking.

42:57

Not enough people spent time in the

42:59

minds of fan fiction.net and it's showing.

43:01

Yeah. Like this is clearly this is

43:03

just like someone wanting to have a

43:06

creative like vision of the future and

43:08

it's like it's not. Interesting or compelling?

43:10

It's joyless. I mean, that's why they

43:12

brought him on. That's why they brought

43:15

Scott Alexander on, to write this narrative,

43:17

right? Because that's what he spends a

43:19

lot of time doing in his blog

43:21

is trying to beautify or flesh out

43:24

why this sort of future is inevitable.

43:26

Yeah. you know, why we need to

43:28

commit to accelerating technological progress as much

43:30

as possible, and why the real reactionary

43:32

or, you know, anti- progress perspective is

43:35

caution or concern or skepticism or criticism,

43:37

if it's not nuanced in a direction

43:39

that supports progress. I just feel like

43:41

a lot of the AI safety guys

43:44

are grifters too. I'm sorry, they love

43:46

saying alignment. Just say pay me. I

43:48

know that we should have, I get

43:50

the occasional email about this being like,

43:53

you can't hate AI safety, it's important.

43:55

It is important. Generative AI isn't AI,

43:57

it's just trying to. I'll fucking accept

43:59

it. If they cared about the safety

44:02

issues, they'd stop burning down zoos and

44:04

feeding entire lakes to generate one busty

44:06

garfield, as I love to say. They

44:08

would also be thinking about the actual

44:11

safety issues of what could this generate,

44:13

which they do. You can't do anarchist cookbook

44:15

shit. It's about as useful. Phil Broughton, friend

44:17

of the show, would be very angry for

44:19

me to bring that up. But the actual

44:21

safety things of it steals from people is

44:23

destroying the environment. It's unprofitable and unsustainable. These

44:26

aren't the actual, these are actual safety issues.

44:28

These are actual problems with this. They don't

44:30

want to solve those. And indeed, the actual

44:32

other safety issue would be, hey, we gave

44:34

a completely unrestrained chatbot to millions of people

44:36

and now they're talking to it like a

44:39

therapist. That's a fucking... That's a safety issue.

44:41

No, they love that. They love it. I

44:43

do think that one criticism of

44:45

the AI safety initiatives that is

44:47

incredibly politically salient and important right

44:49

now is that they are so

44:51

hyper focused on the long-term thousand

44:53

hundred years from now future where

44:55

AI is going to be inside

44:57

all of us and we're all

44:59

going to be, you know, robots

45:01

controlled by and over like that

45:03

they are not paying attention to

45:05

literally any of the harms happening

45:08

right now. do something at work when they

45:10

get aged out. You know, it's like

45:12

when 972 MAG reported on how Israel

45:14

was using or trying to integrate artificial

45:16

intelligence until it generating its killless and

45:18

targets so much so that they started

45:20

targeting civilians and use that to fine-tune

45:22

targeting of civilians. You know I saw

45:24

almost nothing in the immediate aftermath of

45:26

this reporting from the AI safety community

45:28

You know no almost no interest in

45:31

like talking about a very real use

45:33

case where it's being used to murder

45:35

as many civilians as possible Silence you

45:37

know and that's a real short-term concern

45:39

that we should have but that's that would

45:41

require the AI safety people to do something

45:43

and what they do is they get into

45:45

work They're making quarter of a million dollars

45:47

a year they get into work. They load

45:49

slack they load slack they load Twitter and

45:51

that's what they do for being being being

45:53

like By 2028, the AI will have

45:55

fucked my wife. And everyone's like, God

45:57

damn it! No! Not our wives! The

46:00

final frontier. But it is all like,

46:02

they want to talk about 10, 15,

46:04

20 years in the future because if

46:06

they had to talk about it now,

46:09

what would they say? Because I could

46:11

give you AI 2026, which is open

46:13

AI runs into funding issues, can't pay

46:15

Corweave, can't pay Crusoe to build the

46:17

data centers in Abilene, Texas, which requires

46:20

Oracle who have raised debt to fund

46:22

that, to take a bath on that.

46:24

Their stock gets here. Corweve collapses because...

46:26

most of Corby's revenue is now going

46:28

to be open AI. Antropic can't raise

46:31

because the funding climate has got so

46:33

bad. Open AI physically cannot raise in

46:35

2026 because Softbank had to take on

46:37

murderous debt to even raise one round.

46:39

And that's just with like one minute.

46:42

You're getting me excited here. No, no,

46:44

no. Next newsletter, baby, and probably a

46:46

two-part. But that's the thing. They don't

46:48

want to do these because... They get,

46:50

okay, they would claim I'd get framed

46:53

as a skeptic. They also don't want

46:55

to admit the thing in front of

46:57

them, because the thing in front of

46:59

them is so egregiously bad. With crypto,

47:02

it was not that big. Metavas, it

47:04

was not that big. I do like

47:06

that Metaburn like 40 billion dollars, and

47:08

there's Yahoo finance piece about this, just

47:10

on mismanagement. It's just like they should

47:13

get. Metai, they should just change. Oh,

47:15

they should add an eye at the

47:17

end. An eye at the end. It's

47:19

just, if anyone talks about what's actually

47:21

happening today, which is borderline identical what

47:24

was happening a year ago, let's be

47:26

honest, it's April. 2025. April 2024 was

47:28

when I put up my first piece

47:30

being like, hey, this doesn't seem to

47:32

be doing anything different. And it still

47:35

doesn't. Even with reasoning, it's just more...

47:37

You just wait for Q3, agent forces

47:39

going on. Yeah, agent, so agent zero

47:41

is going to come out. Actually, the

47:43

information reported that sales force is not

47:46

having a good time sailing agent force.

47:48

You never guess why. Wow. Turns out

47:50

that it's not that useful due to

47:52

the problems of generative AI. If only

47:54

someone had said something which the information...

47:57

Backed on the information a little bit,

47:59

but they are actually doing insanely good

48:01

work. Like Corey Weinberg... at least a

48:03

car dizzy, Paris of course, but I'm

48:06

specifically talking about the AI, the AI

48:08

team is fantastic. And like, it's great

48:10

because we need this reporting for when

48:12

this just collapsed so that we can

48:14

say what happened. Because it's going to,

48:17

if I'm wrong, and man would that

48:19

be embarrassing, just gonna be honest, like

48:21

if I'm wrong here, I'm gonna look

48:23

like a huge idiot. But if if

48:25

I'm right here like... Everyone has over

48:28

leveraged on one of the dumbest ideas

48:30

of all to it. Like silly, silly.

48:32

It would be like crypto. It would

48:34

be like if everyone said actually crypto

48:36

will replace the US dollar and you

48:39

just saw like the CEO of Shopify

48:41

being like, okay, I'm gonna go buy

48:43

a beer now using crypto. No, this

48:45

is gonna take me 15 minutes. Sorry.

48:47

That's just for you to get the

48:50

money. Actually it's gonna be more like

48:52

20. The network's busy. Okay, well, how's

48:54

your day. Oh, oh, oh, you use

48:56

your day. Oh, you use money. Oh,

48:59

you use money. Oh, you use money.

49:01

Oh, you use money. Oh, you use

49:03

money. Oh, you use money. Oh, you

49:05

use money. Oh, you use money. Oh,

49:07

you use money. Oh, you use money.

49:10

Oh, oh, oh, oh, oh, oh, oh,

49:12

oh, oh, oh, oh, oh, oh, oh,

49:14

oh, oh, oh, oh, oh, oh, But

49:16

it's what we're doing with AI. It's

49:18

like, well, AI is changing everything. How?

49:21

It's a chat book. What do we

49:23

have an Uber scenario where maybe they

49:25

abandoned the dream of like this $3

49:27

trillion dressable market that's worldwide? They abandoned

49:29

the dream of like being monopoly in

49:32

every place and focus on a few

49:34

markets. And some algorithmic price fixing so

49:36

that they can figure out how to

49:38

juice fares as much as possible, reduce

49:40

the wages as much as possible, and

49:43

finally eke out that profit. What if

49:45

we see, you know, some of these

49:47

firms, they pull back on the ambition

49:49

or the scale, but they persist and

49:51

they sustain themselves because they move on

49:54

to some smaller vision. Like Occam's razor,

49:56

that's the most likely situation is that,

49:58

you know, AI tools are useful in

50:00

some way for some slice of people

50:03

and make... a lot of maybe maybe

50:05

let's be optimistic it makes a sick

50:07

a sizable chunk of a lot of

50:09

people's jobs somewhat easier like it's okay

50:11

it was that worth spending billions and

50:14

billions of dollars and also burning down

50:16

a bunch of trees has that happened

50:18

though like I'm just That could be,

50:20

I think, best case scenario. No, I'm

50:22

not saying you're wrong, I'm just saying

50:25

like, we haven't even reached that yet,

50:27

because with Uber, it was this incredibly

50:29

lossy and remains quite a lossy business,

50:31

but still delivers people to and from

50:33

places and objects from places. Yeah, you

50:36

know, you don't have to, you know,

50:38

as much as I hate them, we're

50:40

going to, you know, you don't have

50:42

that, less drunk driving, you know, you

50:44

know, and some transit in parts of

50:47

parts of parts of cities, in parts

50:49

of cities, This is like if Uber,

50:51

if every ride was $30,000 and every

50:53

car weighed 100,000 times. When you factor

50:56

in the externalities, like pollution maybe. But

50:58

that's the crazy thing. I think generative

51:00

AI is so much worse as well,

51:02

pollution wise. But even pulling that back,

51:04

it's like, I think Open AI just

51:07

gets wrapped into... Copilot. I think that

51:09

that's literally they just shut this shit.

51:11

They're like they absorb Sam Altman into

51:13

the hive mind and he be but

51:15

I think there's a my chaos pick

51:18

for everyone is a Saturn Adela is

51:20

fired and Amy hood takes over if

51:22

that happens I think is Prometheus the

51:24

one who can see stuff I'm fucking

51:26

I don't read no he gave fire

51:29

to more or technique I just spared

51:31

fire. It's It's just frustrating. It's frustrating

51:33

as well because a lot of the

51:35

listeners on the show email me and

51:37

they took like teachers of being like

51:40

are they forcing AI in here, librarian,

51:42

oh there's AI being forced out. I

51:44

mean the impact on the educational sector

51:46

especially with public schools, it's really terrifying

51:48

especially because The school districts and schools

51:51

that are being forced to use this

51:53

technology, of course, are never the private

51:55

wealthy schools. It is the most resource-starved

51:57

public schools that are going to have

52:00

budgets for teachers increasingly cut. Meanwhile, they

52:02

do another AI contract and off-source, like

52:04

lesson, the sort of things that these

52:06

companies, the ed tech AI things, pitch

52:08

as their use case is. lesson planning,

52:11

writing, report. cards. Basically, all the things

52:13

that a teacher does other than physically

52:15

being there and teaching, which in some

52:17

cases the companies do that too. They

52:19

say instead of teaching, put your kid

52:22

in front of a laptop and they

52:24

talk to a chatbot for an hour.

52:26

And that's the thing. And the school

52:28

could of course, I don't know, spend

52:30

money on something that's already being spent,

52:33

which is teachers have to buy their

52:35

own fucking supplies all the time. Teachers

52:37

have to just spend a bunch of

52:39

their money on their money on the

52:41

school. They should ban it at universities

52:44

as well. Everything I'm hearing there is

52:46

just like real fucking bad like the

52:48

I mean the issue is from talking

52:50

to university professors. It's like impossible for

52:53

universities to ban it. Can you elaborate?

52:55

I haven't talked to any. I guess.

52:57

Professors are, the obvious example is like

52:59

essays, like professors get AI written essays

53:01

most the time and they can't figure

53:04

out whether they are AI written or

53:06

not. They just notice that all of

53:08

your students seem to be suddenly be

53:10

doing worse in class while having similar

53:12

output of written assignments. There are very

53:15

few tools for them to be able

53:17

to accurately detect this and figure out

53:19

what to do from it. Meanwhile, I

53:21

guess getting involved in trying to like

53:23

prosecute someone for doing this within the

53:26

academic system is a whole other thing.

53:28

But on the, in K through 12,

53:30

especially, it's been kind of, it's been

53:32

especially frustrating to see that some of

53:34

the biggest pushers of AI end up

53:37

being teachers themselves because they are overworked,

53:39

underpaid, have no time to do literally

53:41

anything, and they have to write God

53:43

knows how many lesson plans and IEPs

53:45

for kids with disabilities, and they can't

53:48

do it all. So like, well, why

53:50

don't I just plug this into what's

53:52

essentially a chat GPT wrapper? And that

53:54

results in worse outcomes for everyone, probably.

53:57

So I have some personal experience with

53:59

IEP. I don't think they're doing it

54:01

there, but they're definitely doing it elsewhere.

54:03

That's one of the things that these

54:05

tools often pitch themselves as you can

54:08

create IEPs. I want to put my

54:10

hands around someone. I want to put

54:12

my hands around someone's fucking. Can you

54:14

describe what an IEP is? I forget

54:16

what it stands for. I think it's

54:19

individual education plan. I don't, I might

54:21

be wrong, but that's the bad. It

54:23

is generally the plan that's put for

54:25

a child with special needs, so autism

54:27

being one, it names exactly what it

54:30

names exactly what it is that they

54:32

have to do. like what the teacher's

54:34

goals will be like socio they like

54:36

legally have to do all the things

54:38

in that document like it's a and

54:41

it changes based on the designation they

54:43

get and so like it's different if

54:45

you get there's like an emotional instability

54:47

one I believe and nevertheless there's like

54:50

separate ones and each one is like

54:52

the goals of the where the kid

54:54

is right now where the kid will

54:56

be in the future and so on

54:58

and so forth the idea that someone

55:01

used chat gPT for at least I

55:03

Wow, how disgraceful as well because it's

55:05

all this weird resource allocation done by

55:07

people and I feel like the overarching

55:09

problem as well as it's the people

55:12

making these decisions, putting this stuff in,

55:14

don't do work. It's school administrators that

55:16

don't teach, it's CEOs that don't build

55:18

anything, it's venture capitalists that have an

55:20

inter- interactive with the economy or anyone

55:23

without a Patagonia sweater and decades. And

55:25

it's these, and again these VCs, they're

55:27

investing money based on how they used

55:29

to make money, which was they invest

55:31

in literally anything and then they sold

55:34

it to literally anyone. And that hasn't

55:36

worked for 10 years. Allison, you mentioned

55:38

the thing 2015-ish. That was when things

55:40

stopped being fun. That was actually the

55:42

last time we really saw anything cool.

55:45

It was around the Applewatch era watch

55:47

era. Really the end of the hype

55:49

cycles the real the success ones at

55:51

least they haven't had one since then

55:54

VR AR X R um crypto metiverse

55:56

the indigo and kickstarter era sharing economy

55:58

Yeah, but these all had the same

56:00

problem which was they cost more money

56:02

than they made and they weren't scalable

56:05

and this is the same problem we've

56:07

had. What we may be facing is

56:09

the fact that the tech industry does

56:11

not know how to make companies anymore.

56:13

Like that may actually be the problem.

56:16

Can I add one thing to what

56:18

you said about people who don't work?

56:20

I think there are people in Silicon

56:22

Valley and I don't, I'm going to

56:24

get a million emails about this, but

56:27

there are a lot of Silicon Valley

56:29

men who are white men who don't

56:31

really socialize. Yep. And I think they

56:33

are kind of propagating this technology that

56:35

allows... others to kind of not interact.

56:38

Like so much of ChatGPT is designed

56:40

to like subvert human interactions. Like you're

56:42

not going to go ask your teacher

56:44

or ask a classmate, hey, how do

56:46

we figure this out? You're just going

56:49

to go to the computer? And I

56:51

think that culturally, like I, you know,

56:53

people who grew up with computers, God

56:55

bless. But, you know, we need to.

56:58

also value social interaction and it's interesting

57:00

that there are these, there's this very

57:02

like small group of people often who

57:04

lack social skills, propagating a technology to

57:06

make other people not have social skills.

57:09

I think there's also a class aspect

57:11

to that because... Totally. Particularly with like

57:13

food on the table, but one thing

57:15

I grew up was I don't trust

57:17

any easy fixes Nothing is ever that

57:20

easy is something that I kind of

57:22

a if something seems too good to

57:24

be true to accessible There's usually something

57:26

you're missing about the incentives or the

57:28

actual output So no, I wouldn't trust

57:31

a computer to tell me how to

57:33

fix something because I don't fucking like

57:35

you made that up I like it

57:37

isn't this easy. There's got to be

57:39

a problem problem. This hallucinations Hi,

57:54

I'm Morgan Sun, host of Close All

57:56

tabs from KQEDD, where every week we

57:58

reveal how the. world collides with everyday

58:00

life. You don't know what's true or

58:02

not because you don't know if AI

58:05

was involved in it. So my first

58:07

reaction was, ha ha, this is so

58:09

funny and my next reaction was, wait

58:11

a minute, I'm a journalist, is this

58:13

real? And I think we will see

58:15

it to a streamer, president, maybe within

58:17

our lifetimes. You can find Close All

58:19

tabs wherever you listen to podcasts. So,

58:21

we didn't really lead into that ad

58:23

break, but you're going to just have

58:25

to like it. I'm sure all of

58:27

you are going to send me little

58:29

emails, little emails about the ads that

58:31

you love. Well, I've got to pay

58:33

for my diet Coke somehow. So, back

58:35

to this larger point around chat-GPT and

58:37

why people use it and how people

58:39

use it. I think another thing that

58:41

just occurred to me is... Have you

58:43

ever noticed that Sam Altman can't tell

58:46

you how it works and what it

58:48

does? You haven't noticed none of these

58:50

people will tell you what it does.

58:52

I've read everything Sam at this point,

58:54

listening to hours of podcasts, he's quite

58:56

a boring twirp. But on top of

58:58

that, for all is the yapping and

59:00

yamoring, him and Warrio Amadeh, don't seem

59:02

to be able to say out loud

59:04

what the fucking thing does. And that's

59:06

because I don't think that they use

59:08

it either. Like I genuinely, I'm beginning

59:10

to wonder if any of the people

59:12

injecting AI. Sure, Sam Altman and Dario

59:14

probably use it. I'm not saying it

59:16

fully, but like, these aren't people, how,

59:18

the next person that meets Sam Altman

59:20

should just be like, hey, how often

59:22

do you use chat cheapity? Gets back

59:24

to that it reminds me of the

59:26

remote work thing all these CEO saying

59:29

guys should come back to the office

59:31

How often you in the office exactly

59:33

and I think that this is just

59:35

the giant revelation of like how many

59:37

people don't actually interact with their businesses

59:39

that don't interact with other people that

59:41

don't really know how anything works But

59:43

they are the ones making the money

59:45

in power decisions. It's fucking crazy to

59:47

me and I don't know how this

59:49

shakes out. It's not going to be

59:51

an autonomous agent doing whatever also okay

59:53

that just occurred to me as well

59:55

How the fuck do these people not

59:57

think these agents come for them first?

59:59

If the AGI was doing this and

1:00:01

they read... this and be like these

1:00:03

people fucking they worked it all out

1:00:05

I need to kill them first Well,

1:00:07

I mean, that kind of gets back

1:00:10

to what you're saying, where it's like,

1:00:12

you know, if we entertain the fan

1:00:14

fiction for a little bit, what is

1:00:16

the frame of mind for these agents

1:00:18

if they're autonomous or not? How are

1:00:20

we thinking of them? Are we thinking

1:00:22

about like, if they're persons or if

1:00:24

they're, you know, the bottomized in some

1:00:26

way? Do they have opinions? You know,

1:00:28

and I think really just gets back

1:00:30

to like, you know, part of the

1:00:32

old hunt for like, you know, you

1:00:34

know, a nice. a polite slave, you

1:00:36

know? How do we figure out how

1:00:38

to reify that relationship? Because it was

1:00:40

quite profitable at the turn of like

1:00:42

industrial capitalism and you know, I think,

1:00:44

you know, it's not a coincidence that

1:00:46

a good chunk of our tech visions

1:00:48

come to us from reactionaries who think

1:00:50

that the problem with capitalism, the problem

1:00:53

with tech development is that a lot

1:00:55

of these empathetic egalitarian reforms get in

1:00:57

the way of profit making. You know,

1:00:59

I think similarly, you know, the hunt

1:01:01

for automatons for certain algorithmic systems is

1:01:03

searching for a way to figure out

1:01:05

how do we replicate, you know, human

1:01:07

labor without the limitations on extracting and

1:01:09

pushing and coercing as much as possible

1:01:11

with, you know, your agent or something

1:01:13

else. And the thing is, yeah, sure,

1:01:15

the idea of an autonomous AI system

1:01:17

would be really useful. I'm sure it

1:01:19

could do stuff. That sounds great. They're

1:01:21

a massive, as you've mentioned, like... Soci

1:01:23

Sociological problems, like, do these things feel

1:01:25

pain? If so, how do I create?

1:01:27

Anyway, but in all seriousness, like, sure,

1:01:29

an autonomous thing that could do all

1:01:31

this thing would be useful. They don't

1:01:34

seem to even speak to that. It's

1:01:36

just like, and then the AI will

1:01:38

make good decisions. And then the decisions

1:01:40

will be even better, then Agent 7

1:01:42

comes out, and you thought Agent 6

1:01:44

was good. It's like they don't even

1:01:46

speak to how we're going to get

1:01:48

to the point where Agent 1 knows

1:01:50

truth from falsehood from falsehood. Of course,

1:01:52

yeah, you know, we just need to

1:01:54

give it all of our data and

1:01:56

everything that we've paid money for required

1:01:58

other people to pay money for. and

1:02:00

then it will finally be perfect. And

1:02:02

it doesn't even make profit of

1:02:05

any kind. That's the other thing.

1:02:07

It's like people saying, well, it

1:02:09

makes profit. It is the profit-seeking.

1:02:12

Is it profit-seeking? It doesn't seem

1:02:14

like we've sought much profit

1:02:16

or any. That's also, I

1:02:18

think, a good point of

1:02:20

comparison to Uber. These companies

1:02:22

that achieved massive scale and

1:02:25

popularity, by making their products

1:02:27

purposefully unprofitable by... charging you $5

1:02:29

for a 30-minute Uber across town so

1:02:31

that you're like, yeah, this is going

1:02:33

to be part of my daily routine.

1:02:35

And the only way they've been able

1:02:37

to squeeze out a little bit of

1:02:39

profit right now is by hiking those

1:02:41

prices up, but trying to balance it

1:02:43

to where they don't hike it up

1:02:45

so much that people don't use it

1:02:47

anymore. And AI is at the point

1:02:49

where for these agents, I think some

1:02:52

of the cost are all something like

1:02:54

thousands of dollars a month. And they

1:02:56

don't work already. you're still not making

1:02:58

money by charging people that much money to

1:03:00

use it. What is the use case one

1:03:02

where this even works? And if it somehow

1:03:04

did manage to work, how much is that

1:03:06

going to cost? Who is going to be

1:03:08

paying $20,000 a month for one of these

1:03:10

things? And how much of that is dependent

1:03:12

on what is clearly nakedly subsidized compute

1:03:15

prices? How much of this is because

1:03:17

Microsoft's not making a profit on a

1:03:19

zero compute? Open AI isn't making anthropicism.

1:03:21

What happens if they need to they

1:03:23

need to. What if they need to?

1:03:26

They're going to, that's the subprime AI

1:03:28

crisis from last year. It's just, it's,

1:03:30

it's- Well, that's when you get the

1:03:32

venture capitalist insisting that that's why we need

1:03:34

to, you know, do this Air CapX rollout,

1:03:37

because if we build it out, like infrastructure,

1:03:39

then we can actually lower the compute

1:03:41

prices and not subsidize anymore? Yeah, that's a

1:03:43

thing. But that's the other thing. So, the

1:03:46

information reported, the Open, Open AI says

1:03:48

that by 2030, they'll be profitable, they'll be

1:03:50

profitable, they'll be profitable, how, how? And you may

1:03:52

think, what does that mean? And the answer is,

1:03:54

Stargate has data centers. Now, you have to, I

1:03:56

just have one little question. This isn't a knock

1:03:59

on the information, this is... they're reporting what

1:04:01

they've been told, which is fine. A

1:04:03

little question with Open AI though, how,

1:04:05

how does more equal less cost? Because

1:04:08

this thing doesn't scale, they lose money

1:04:10

on every prompt, it doesn't feel like

1:04:12

they'll make any money, in fact they

1:04:15

won't make any money, they'll just have

1:04:17

more of it. And also there's the

1:04:19

other thing of... Data centers are not

1:04:22

fucking weeds, they don't grow in six

1:04:24

weeks, they take three to six years

1:04:26

to be fully done. If Stargate is

1:04:29

done by next year, I will fucking

1:04:31

barbecue up my parjorie's hat and eat

1:04:33

it live on stream. Like, I will,

1:04:35

that's if they're fucking alive next year.

1:04:38

Also, the other thing is, getting back

1:04:40

to 2027 as well, year 2026-2027 is

1:04:42

gonna be real important for everything. 2027

1:04:45

or 2026 is when Warrio Abedabadadee says

1:04:47

that Anthropable be profitable be profitable to

1:04:49

be profitable. That's also when the Stargate

1:04:52

Data Center project will be done in

1:04:54

2026. I think that they may have

1:04:56

all just chosen the same year because

1:04:59

it sounded good and they're going to

1:05:01

get in real trouble next year when

1:05:03

it arrives and they're nowhere near close.

1:05:06

I can't wait until all of those

1:05:08

companies announce. that because of the tariffs,

1:05:10

that they have to delay their timeline

1:05:13

and it's like completely out of their

1:05:15

hands, but no, the tariffs, you understand

1:05:17

the tariffs. I got a full roasted

1:05:20

pig, I'm going to be tailgating. Microsoft

1:05:22

earnings, April 23rd, cannot wait. Yeah, you

1:05:24

should go to like a data center.

1:05:27

No, no, no, no. You have a

1:05:29

marching band, all the end of severance.

1:05:31

Yeah, it's, but that's the thing. Like,

1:05:33

I actually agree. I think that they're

1:05:36

going to, there's going to be difficult

1:05:38

choices. Sadly, there's only really two. One,

1:05:40

CapX reduction, two layoffs. Like why are

1:05:43

we doing this? It just, it feels

1:05:45

like the collapse of any good or

1:05:47

bad romantic relationship, where just one party

1:05:50

is doing shit that they think works

1:05:52

from years ago and the other party

1:05:54

is just deeply unhappy and then disappears

1:05:57

one day. and the other party being

1:05:59

just happened. I watched an episode last

1:06:01

last night and it just happened. This

1:06:04

is, uh... Lost is a far more

1:06:06

logical show than any of this AI

1:06:08

bullshit, but it's... Let's not get that

1:06:11

crazy. No, no, it's a bad show.

1:06:13

It's about... No, I wouldn't say that

1:06:15

either. Talking about something that's very long,

1:06:18

very expensive and never had a plan,

1:06:20

but everyone talks about like it was

1:06:22

good, despite and never proving it, lost.

1:06:25

Yeah, sorry I have some I really

1:06:27

do have some feelings on that for

1:06:29

you're gonna get you're gonna get some

1:06:31

emails I I'm sure I email for

1:06:34

me Yeah He's texting me it is

1:06:36

writing and it is just sending me

1:06:38

the very quiet right crystal emotion Emogy

1:06:41

like a hundred times. Yeah, it's I

1:06:43

think it's just I can't wait to

1:06:45

see how people react to this stuff

1:06:48

as well when this because I obviously

1:06:50

will look very silly of these companies

1:06:52

stay alive and somehow make a KGI

1:06:55

Like the gravedigger AI truck is going

1:06:57

to run me over outside my house,

1:06:59

it's going to be great. But I

1:07:02

can't wait to see how people explain

1:07:04

this. I can't wait to see what

1:07:06

the ex, it's like, are we never,

1:07:09

the tariffs maybe? Right, and I talked

1:07:11

to an analyst just last week who's

1:07:13

like a bullish AI tech investor and

1:07:16

he said, he said already you're seeing

1:07:18

investment pull back because of expectations in

1:07:20

the market that there was... these stocks

1:07:23

were overbought in the first place and

1:07:25

now there's all this other turmoil external

1:07:27

macro elements that are going to kind

1:07:29

of take the you know the jargon

1:07:32

of like the froth out of the

1:07:34

market they're gonna it's all gonna deflate

1:07:36

a little bit and so I was

1:07:39

asking him like is the AI bubble

1:07:41

popping and he says no but tariffs

1:07:43

are definitely like deflating it and and

1:07:46

delaying whatever progress that we are going

1:07:48

to be promised from these companies is

1:07:50

going to be delayed. even if it

1:07:53

was going to be delayed they were

1:07:55

going to find other reasons this is

1:07:57

a convenient macro kind of excuse to

1:08:00

just say like oh well we need

1:08:02

we didn't have enough chips we didn't

1:08:04

have enough investing we didn't have enough

1:08:07

you know, be patient with us. We're

1:08:09

going to have the revolution is coming.

1:08:11

What's great is well, talking of my

1:08:14

favorite Wall Street analyst, Jim Kramer of

1:08:16

CNBC. So Corweaves IPO went out. I

1:08:18

just need to mention we are definitely

1:08:21

in the hype cycle because Jim Kramer

1:08:23

said that he would sue an analyst,

1:08:25

DA Davidson, on behalf of Invidia, for

1:08:27

claiming that they were a lazy Susan,

1:08:30

as in... As in, basically what the

1:08:32

argument is, is the invidia-funded coreweave, so

1:08:34

the coreweave would buy GPUs, and at

1:08:37

that point, coreweave would then take out

1:08:39

loans on those GPUs for cap-ex reasons,

1:08:41

cap-ex, including buying GPUs. So very clearly,

1:08:44

and also you tack gill over at

1:08:46

the A. Davison, you and me, Kramer,

1:08:48

in the ring. But we know we're

1:08:51

in the crazy time when you've got

1:08:53

like a TV show host being like

1:08:55

a TV show host being like a

1:08:58

TV show. I think that we're going

1:09:00

to see a historic washout of people

1:09:02

and the way to change things is

1:09:05

this time we need to make fun

1:09:07

of them. I think we need to

1:09:09

be like actively, we don't need to

1:09:12

be mean, that's my job, but we

1:09:14

can be like, to your point, your

1:09:16

article. Allison, it's like, say like, hey

1:09:19

look, no, what you are saying is

1:09:21

not even rational or even connected to

1:09:23

reality. This is not doing the right

1:09:25

things. Apple intelligence is like the greatest...

1:09:28

Anti-A-A-A-I radicalization ever. I actually think Tim...

1:09:30

It's so bad. It's so fucking bad.

1:09:32

And I, before it even came out,

1:09:35

I like downloaded the beta, I was

1:09:37

like, I'm going to test this out

1:09:39

because, you know, I talk about this

1:09:42

thing on my podcast sometimes, and it's

1:09:44

so bad. It's so bad. I like

1:09:46

turns it off for most things, but

1:09:49

I have it on for a couple

1:09:51

of social networks. And I mean, I

1:09:53

guess with the most recent recent recent

1:09:56

update. I check they didn't that person

1:09:58

didn't even like the ski. I don't

1:10:00

know where that name came from and

1:10:03

this happens like every other day. It's

1:10:05

just completely wrong. I'm like how? My

1:10:07

favorite is the summary text for Uber.

1:10:10

where it's like several cars headed to

1:10:12

your location. It's great as well because

1:10:14

I usually don't buy into the Steve

1:10:17

Jobs would burst from his grave thing.

1:10:19

I actually think numerous choices Tim Cook

1:10:21

has made have been. way smart on

1:10:23

how jobs would run. This is actually

1:10:26

like he's going to burst out of

1:10:28

the ground, thriller style, actually did, was

1:10:30

that zombies pop out? Anyway, because it's

1:10:33

nakedly bad, close, it's not a great

1:10:35

reference, but it's nakedly bad, like it

1:10:37

sucks, and I've never, I've, people in

1:10:40

my life were non-techie, constantly be like,

1:10:42

hey, what is Apple intelligence? Am I

1:10:44

missing something? I'm like, no, it's actually

1:10:47

as bad as you think. word and

1:10:49

I'm trying to sometimes might want to

1:10:51

use fine definition or any of the

1:10:54

things that come up I have to

1:10:56

scroll by like seven different new options

1:10:58

under the like right click or double

1:11:01

click thing. And if you hit writing

1:11:03

tools it opens up a screen white

1:11:05

thing. Yes, it opens up a thing

1:11:08

and I'm like who has ever who

1:11:10

is trying to use this to rewrite

1:11:12

a text to their group chat? Who

1:11:15

is this for? I feel like Apple

1:11:17

to its credit is recognizing its mistake

1:11:19

and it's clawing it back and like

1:11:21

delaying Syria indefinitely. I mean... I don't

1:11:24

know if I agree on that one.

1:11:26

That's fair. Because the thing they're delaying

1:11:28

is the thing that everyone wanted. I

1:11:31

think they can't make it work because

1:11:33

the thing they're delaying is never existed.

1:11:35

Yeah. We'll see. Apple's washed. I mean,

1:11:38

but that's the thing. It's the most

1:11:40

brand conscious company on the planet. And

1:11:42

I wrote like when they did their

1:11:45

June revelation of the Syria I was

1:11:47

going to come out and they said

1:11:49

it was going to come out in

1:11:52

the fall and then it was coming

1:11:54

out in the spring and now it's

1:11:56

not coming out ever, question mark. about

1:11:59

the whole like two-hour presentation. The letters

1:12:01

AI were never spoken, artificial was never

1:12:03

spoken, it was Apple intelligence. We're doing

1:12:06

this, we're doing our own thing, it's

1:12:08

not, you know, because they already understood

1:12:10

that when you say something is like,

1:12:13

that looks like it was generated by

1:12:15

AI, you're saying it looks like shit,

1:12:17

you know. And the suggestions are also

1:12:19

really bad too. I've had like over

1:12:22

the last few weeks a few people

1:12:24

give me some bad news from their

1:12:26

lives and the responses it gives are

1:12:29

really funny. Oh no. It'd be like

1:12:31

someone telling me something bad happened and

1:12:33

it's like, oh, or like I'm like,

1:12:36

what was the worst one I had?

1:12:38

It was like, that sounds difficult and

1:12:40

it's like, like, any juice to it,

1:12:43

like. I didn't read too long. Those

1:12:45

would be funny suggestions. But it's proof

1:12:47

that I think that these large language

1:12:50

models don't actually, well they don't understand

1:12:52

anything, they don't know, I think they're

1:12:54

not conscious. But it's like, they're really

1:12:57

bad at understanding words. Like people like,

1:12:59

oh, they make some mistakes. They're bad

1:13:01

at basic contextual stuff. And we had

1:13:04

Victoria song from The Virgin on the

1:13:06

other day, and she was talking about

1:13:08

high context and low context languages languages.

1:13:11

I can only speak English. I imagine,

1:13:13

not being able to read or speak

1:13:15

in any others, that it really fumbles

1:13:17

those. And if you're listening, you want

1:13:20

to email me anything about this research,

1:13:22

how the fuck does this even scale

1:13:24

if it can't, oh we're replacing translators,

1:13:27

great, you're replacing translators with things that

1:13:29

sometimes translate, right? Sometimes? It just feels,

1:13:31

it also inherently, like that feels like

1:13:34

an actual. alignment problem by the way

1:13:36

that right there that feels like an

1:13:38

actual safety problem okay if we're relying

1:13:41

on something to translate and it translates

1:13:43

words wrong and you know especially in

1:13:45

other languages subtle differences can change everything

1:13:48

maybe that's dangerous no no no we

1:13:50

got the computer wake up in like

1:13:52

two weeks and then it's gonna be

1:13:55

And that's the other thing. We're going

1:13:57

to make AGI and we think it's

1:13:59

not going to be pissed off at

1:14:02

us. I don't mean Rokoko's modern basilisk

1:14:04

or whatever. I mean just like if

1:14:06

it wakes up and looks at the

1:14:09

world and goes, these fucking morons, like

1:14:11

you need to watch person of interest

1:14:13

if you haven't one of the best

1:14:16

shows on actually on AGI, like genuinely

1:14:18

you need to watch person of interest

1:14:20

because you will see how that could

1:14:22

happen when you allow a quote perfect

1:14:25

computer to be in particularly good at...

1:14:27

decision-making? I don't know. I feel like

1:14:29

so much of this revolution, quote-unquote, is

1:14:32

based on just the assumption that the

1:14:34

computer makes great decisions and it oftentimes

1:14:36

doesn't. It often does not. Why would

1:14:39

I think that the same search function

1:14:41

in Apple that cannot find a document

1:14:43

that I know what the name is

1:14:46

and I'm searching for it? Why would

1:14:48

I think that that same computer is

1:14:50

going to be able to make wise

1:14:53

decisions about my life finance and personal

1:14:55

relationships? Because that's Apple and this is

1:14:57

AI. Oh, that's, that's, I'll show myself

1:15:00

out. I don't know how much is

1:15:02

AI versus just like a good translation

1:15:04

app. Like I genuinely don't know like

1:15:07

well it's because AI is such a

1:15:09

squishy term that we really don't like

1:15:11

right in some way like I guess

1:15:14

AI could be expanded to include a

1:15:16

lot of modern computing like I can

1:15:18

see travel in like emergency situations where

1:15:20

you need where like a good AI

1:15:23

translator would be like a real lifesaver

1:15:25

just as a small aside I was

1:15:27

just in Mexico and my stuff kids

1:15:30

were using Google Translate and it were

1:15:32

like kind of remembering Spanish and you

1:15:34

know blah blah blah. Go into a

1:15:37

coffee shop and I wanted to order

1:15:39

a flat white and so I used

1:15:41

Google Translate to say like how would

1:15:44

you order a flat white in Spanish

1:15:46

and it said to order a Blanco

1:15:48

Plano which means flat white but like

1:15:51

across Mexico City there are wonderful coffee

1:15:53

shops and you know what they call

1:15:55

them? Flat whites. and Australian coffee or

1:15:58

something? I learned that very quickly with

1:16:00

the help of Reddit, because I went

1:16:02

to the barista and ordered a Blanca

1:16:05

Plano and they were like, who are

1:16:07

you? You crazy Grinko can have here.

1:16:09

Sorry, I speak English. Yeah. Yeah. I

1:16:12

mean, the functionality is very limited on

1:16:14

those things. And it's just like, also,

1:16:16

it gets back to like, if it's

1:16:18

100% reliable, it's great. If it's 98%

1:16:21

reliable, it sucks. And, um, just as

1:16:23

an aside, did any of you hear

1:16:25

about Joan, like, the latest, like, quasi-fraudulent

1:16:28

thing with Joanie Ive that's happening? I

1:16:30

just saw the headline. So, Sam Altman

1:16:32

and Joanie Ive founded a hardware star

1:16:35

of last year. No, actually, I do

1:16:37

know. That has built nothing. There is

1:16:39

a thing they claim there's a phone

1:16:42

without a screen. Sick. And Open AI,

1:16:44

a company run by Sam. for instance,

1:16:46

by Sam Altman and Microsoft, is going

1:16:49

to buy for half a billion dollars

1:16:51

this company that has built nothing, co-founded

1:16:53

by Sam Altman. Sick. I feel like

1:16:56

there should be a one law against

1:16:58

this, but it's just like, what have

1:17:00

they been doing? And this is just,

1:17:03

it's kind of cliched to say like,

1:17:05

quote the big short, but like a

1:17:07

big part of the beginning of that

1:17:10

movie is talking about the increase in

1:17:12

fraud and fraud and scams. And it

1:17:14

really feels like we're getting there. Rest

1:17:16

in piss, you won't be missed. Motherfuckers,

1:17:19

two management consultants, both like in dignity,

1:17:21

each shit. Jesse Lou, Jesse Lou, rabbit,

1:17:23

oh one, you're next, motherfucker. When your

1:17:26

shit's gone, I'll be honking and laughing.

1:17:28

Your customers should suit you. The description,

1:17:30

so my colleagues, the information reported this

1:17:33

joining I of Sam Altman News, and

1:17:35

the description for the device really makes

1:17:37

me chuckle. Designs for the AI device

1:17:40

are still early and haven't been finalized,

1:17:42

the people said. Potential designs include a

1:17:44

quote-unquote phone without a screen, and AI-enabled

1:17:47

household devices. Others. close to the project

1:17:49

are adamant that it is, quote, not

1:17:51

a phone, end quote. And they're going

1:17:54

to spend, they've discussed spending upwards of

1:17:56

500 million on this company. This is

1:17:58

like a bad philosophy class where it's

1:18:01

like, what is a phone that's not

1:18:03

a phone? Lipot and phone. Semiotics for

1:18:05

beginners. Jesus, fucking Christ. Oh my God.

1:18:08

And that's, like, that's the thing as

1:18:10

well, like, this feels like a thing

1:18:12

the tech media needs to be on.

1:18:14

Just someone needs to say, I'll be

1:18:17

saying it. Boardering on fraud, like it

1:18:19

seems like it must be legal because

1:18:21

otherwise there would be some sort of

1:18:24

authority, right? You can't do anything illegal

1:18:26

without anything happening. Hmm. But it's like,

1:18:28

this is one of the most egregious

1:18:31

fucking things I've ever seen. This is

1:18:33

a guy handing himself money. One hand,

1:18:35

this is, should be fraud. Like this,

1:18:38

how is this ethical? And everyone's just

1:18:40

like, oh yeah, you know. This, Kevin

1:18:42

Roose. Maybe you should get on this

1:18:45

shit, find out what the phone that

1:18:47

isn't the phone is. What the fuck?

1:18:49

And also household appliances with AI. Maybe

1:18:52

like something with the screen and the

1:18:54

speaker that you could say, like a

1:18:56

word to it and it would wake

1:18:59

up and play music. Like a rumba.

1:19:01

A rumba? A rumba with AI mic.

1:19:03

Just declared bankruptcy. D. J. Rumba dead?

1:19:06

Rumba dead? Why rumba dead? I think

1:19:08

they did, I don't know actually, I

1:19:10

remember, I read the headline in the

1:19:12

last few weeks. They were supposed to

1:19:15

be acquired by Amazon, but I think

1:19:17

the deal fell through under Lina Khan's

1:19:19

FTC, I'd assume? Sick. Got them. Hmm.

1:19:22

Also a one quick note on the

1:19:24

joining I have Sam Multan thing. I

1:19:26

guess it's notable that Aldman has been

1:19:29

working closely with the product but is

1:19:31

not a co-founder and whether he has

1:19:33

an economic stake in the hardware project

1:19:36

is unclear. Yeah, you know, he just

1:19:38

seems to be working closely with it.

1:19:40

He just ends up with a room.

1:19:43

He's just hanging out there on taking

1:19:45

a salary and an equity position. I

1:19:47

do think it's very interesting. All of

1:19:50

these different AI device startups that have

1:19:52

popped up in the last couple of

1:19:54

years. question for them is always just

1:19:57

like To what end? Yeah. People didn't

1:19:59

like Amazon Alexa. And it also lost

1:20:01

a shoot ton of money. Yeah. And

1:20:04

Amazon's still trying to make it work.

1:20:06

Series never been super popular. And I

1:20:08

just don't, like, one of my co-hosts

1:20:10

on the podcast for intelligent machines is

1:20:13

obsessed with all these devices, just because

1:20:15

he's like one of those tech guys.

1:20:17

Leo, yes. He's, and we love to

1:20:20

make fun of it. He his latest

1:20:22

device is this thing called a B.

1:20:24

We just have Victoria's song on talking

1:20:27

about this the thing that records everything

1:20:29

all the time and then Makes puts

1:20:31

that up in the cloud and then

1:20:34

I guess doesn't store the full transcripts

1:20:36

But does store a little AI-generated description

1:20:38

of everything you did and whoever you

1:20:41

talked to that day and there's no

1:20:43

way I mean he's in California. Which

1:20:45

is not a one-party recording. Yeah everybody

1:20:48

to record and the B is not

1:20:50

doing that but it's just baffling to

1:20:52

me because I'm just like I guess

1:20:55

he's like well it could be nice

1:20:57

to have record of all of my

1:20:59

days all the time and I'm like

1:21:02

I guess but to what end just

1:21:04

record it you go record it down

1:21:06

yeah right it down there's literally a

1:21:08

black mirror episode about that I believe

1:21:11

it's the first episode of black mirror

1:21:13

everyone has like a recording device and

1:21:15

then it does when I was when

1:21:18

you were talking about this on the

1:21:20

show I was listening thinking like in

1:21:22

this black mirror thing it reminded me

1:21:25

that like when Facebook started having like

1:21:27

all your photos collected under your photos

1:21:29

and like how we started reliving so

1:21:32

many experiences on mine like you could

1:21:34

scroll back and like look at how

1:21:36

happy you were like six years ago

1:21:39

you know like and it creates this

1:21:41

like cycle like imagine if every interaction

1:21:43

every like romantic interaction every sad inner

1:21:46

everything you could replay back to yourself.

1:21:48

It's sad. like a nightmare to me.

1:21:50

I do think it's also just a

1:21:53

night, like humans, we're not built socially

1:21:55

to exist in a world where every

1:21:57

interaction is recorded and searchable with everyone

1:22:00

forever. Like you would never have a

1:22:02

single friend. Romantic relationships would dissolve. The

1:22:04

eternal sunshine on the spotless line, like,

1:22:06

maybe, frame, but even then, like, memory

1:22:09

is vastly different to the experience of

1:22:11

collecting it. Like just existing, like we

1:22:13

are brain slot, I don't know, my

1:22:16

brain just goes everywhere. But like compared

1:22:18

to memory which can be oddly crystalline

1:22:20

and wrong, you can just remember something,

1:22:23

you can remember a subtle detail wrong,

1:22:25

or you can just fill in the

1:22:27

gaps, memory sucks. Also doesn't having like

1:22:30

a device that constantly records everything? a

1:22:32

road at the impulse or maybe the

1:22:34

drive to be as present. You know,

1:22:37

because you're like, well, it's going to

1:22:39

refer to it. But this is also

1:22:41

got huge privacy implications where suddenly the

1:22:44

cops could just be like, yeah, we're

1:22:46

just going to take a recording, we're

1:22:48

just going to subpoena everybody who was

1:22:51

in this area's B device, and then

1:22:53

suddenly get a recording of everyone's days

1:22:55

ever that was just happened to be

1:22:58

in this place because we think a

1:23:00

crime could have happened there. But I

1:23:02

think that there's an overarching thing to

1:23:04

everything we're talking about which is these

1:23:07

are products made by people that haven't

1:23:09

made anything useful in a while and

1:23:11

everything is being funded based on what

1:23:14

used to work. What used to work

1:23:16

was you make a thing, people buy

1:23:18

it and then you sell it to

1:23:21

someone else to take it public. This

1:23:23

only worked until about 2015. It's not

1:23:25

just a zero interest free everything. We

1:23:28

have increasingly taken away. the creation of

1:23:30

anything valuable in tech from people who

1:23:32

experience real life. Like our biggest CEOs

1:23:35

are Sam Altman, Wario Amadee, Sundar Pishai,

1:23:37

MBA, former McKinsey, Sachin Adele of MBA.

1:23:39

I mean, Tim Cook, MBA. Like these

1:23:42

are all people that don't really interact

1:23:44

with people anymore and the problems, the

1:23:46

people in power are not engineers, they're

1:23:49

not... Even startup founders anymore, they're fucking

1:23:51

business people. Making things that they think

1:23:53

they could sell, things that could grow

1:23:56

the right economy, of course. And we're

1:23:58

at the kind of the pornographic point

1:24:00

where it's like a guy being like

1:24:02

what could a what is a what

1:24:05

is a I do you can just

1:24:07

throw a bunch of data and give

1:24:09

you insights well what if we just

1:24:12

collected date on everything happening around us

1:24:14

ever that would be good then you

1:24:16

could reflect on things that's what people

1:24:19

do right and I actually genuinely think

1:24:21

there was only one question to us

1:24:23

to be found on that so are

1:24:26

you wearing one of these now and

1:24:28

how often do you use this because

1:24:30

like if they use it all the

1:24:33

kind of respect them I guarantee they

1:24:35

don't. I guarantee they don't, and they'll

1:24:37

probably say something called around a lot

1:24:40

of privileged information, as opposed to everyone

1:24:42

else who's not important. And there's fucking

1:24:44

Joanie, oh, it's going to be a

1:24:47

phone without a screen. What can you

1:24:49

do with it? I don't know. I

1:24:51

haven't thought that far ahead. I only

1:24:54

get paid $15 million a year to

1:24:56

do this. Question is also, who wants

1:24:58

a phone about a screen? The screen's

1:25:00

the best part. I love the screen.

1:25:03

I love to hate- The screen. I

1:25:05

love to hate to hate to hate

1:25:07

to hate to hate to hate to

1:25:10

hate to hate to hate- the like

1:25:12

they have friends who are all like

1:25:14

have fifty million dollars in the bank

1:25:17

account at any time they just like

1:25:19

exist in a different difficulty level they're

1:25:21

all going at very easy they don't

1:25:24

really have like predators of any kind

1:25:26

they don't really have experiences so they're

1:25:28

what they experience in life is when

1:25:31

you have to work out what you

1:25:33

enjoy and because they enjoy nothing all

1:25:35

they can do is come up with

1:25:38

ideas that's why the rabbit are one

1:25:40

oh what do people do uh uh...

1:25:42

autumn McDonald's can it do it do

1:25:45

it not really But it also could

1:25:47

take a photo of something, it could

1:25:49

be pixelated. That could all. You could

1:25:52

kind of order an Uber through it,

1:25:54

maybe. What was great was the rabbit

1:25:56

launch, the rabbit launch, and he tried

1:25:58

to order McDonald's live, and it just

1:26:01

didn't work. Took like five minutes to

1:26:03

fail. And that's the thing, like, I

1:26:05

feel like when this hype cycle ends,

1:26:08

the tech media needs to just be

1:26:10

aggressively like, we can ask the questions

1:26:12

I was asking in 2021 where it's

1:26:15

like, what does this do? Who is

1:26:17

it for? And if anyone says it.

1:26:19

could address millions of papers. Like, have

1:26:22

you talked to one of them, motherfucker?

1:26:24

One of them. Well, I think we

1:26:26

can wrap it up there, though. I

1:26:29

think, Allison, where can people find you?

1:26:31

You can find me at cnn.com/Nightcap. I

1:26:33

write the CNN business Nightcap. It's in

1:26:36

your inbox four nights a week. Oh,

1:26:38

yeah. Ed? I write a newsletter on

1:26:40

Substack Tech Bubble. I co-host podcast This

1:26:43

Machine Machine Machine Machine Kills. You know,

1:26:45

Blue Sky to, right? Yeah, Blue Sky

1:26:47

at Edward Unguesso, junior.com. You can read

1:26:50

my work at the information. I also...

1:26:52

host a podcast called Intelligent Machines and

1:26:54

you can find me on Twitter at

1:26:57

Paris Martino or on Blue Sky at

1:26:59

Paris.NYc. And you can find me at

1:27:01

at edzitron.com on Blue Sky Google who

1:27:03

destroyed Google search click the first link

1:27:06

it's me I destroyed Google search click

1:27:08

the first link it's me I destroyed

1:27:10

Google search along with pobago ragabam fuck

1:27:13

you dude if you want to support

1:27:15

this podcast you should go to the

1:27:17

webbies I will be putting the link

1:27:20

in there I need your help I've

1:27:22

never won an award in my award

1:27:24

in my life We are winning right

1:27:27

now, please help me, please help me

1:27:29

win this, and if I need to

1:27:31

incentivize you further, we are beating Scott

1:27:34

Galloway. If you want to defeat Scott

1:27:36

Galloway, you need to vote on this.

1:27:38

Thank you so much for listening everyone.

1:27:41

Thank you for listening to Better Offline.

1:27:43

The editor and composer of the Better

1:27:45

Offline theme song is Matosowski. You can

1:27:48

check out more of his music and

1:27:50

audio projects at matosowski.com. m-a-t-t-o-s-s-o-w-s-s-k-i.com. You can

1:27:52

email me at easy at betrooffline.com or

1:27:55

visit betrooffline.com to find more podcast links

1:27:57

and of course my newsletter. Thank you

1:27:59

so much for listening. Better offline is

1:28:01

a production of Cool Zone Media. For

1:28:04

more from Cool Zone Media, visit our

1:28:06

website, coolzonemedia.com, or check us out on

1:28:08

the I heart radio app, Apple Podcast,

1:28:11

or wherever you get your podcast. Military

1:28:35

veterans and active duty service

1:28:37

members. You could become a

1:28:39

certified cyber warrior with just

1:28:41

12 weeks of training at

1:28:43

my computer career. Cybersecurity specialists

1:28:46

are in high demand. And

1:28:48

IT pros with these skills

1:28:50

often enjoy abundant opportunity and

1:28:52

a great lifestyle. Protect our

1:28:54

people, liberty, and treasured institutions

1:28:56

from cyber threats. Skillbridge and

1:28:58

other VA benefits are available

1:29:00

to those who qualify. Learn

1:29:02

more at my computercareer dot

1:29:04

EDU slash CWP. In

1:29:06

a world of economic uncertainty and workplace

1:29:08

transformation, learn to lead by example from

1:29:11

visionary C-sweet executives like Shannon Skyler of

1:29:13

PWC and Will Pearson of Ihart Media.

1:29:15

The good teacher explains, the great teacher

1:29:17

inspires. Don't always leave your team to

1:29:20

do the work. That's been the most

1:29:22

important part of how to lead by

1:29:24

example. Listen to Leading by Example, executives

1:29:26

making an impact on the I heart

1:29:29

radio app, Apple Podcasts, or wherever you

1:29:31

get your podcasts. Hi, I'm Morgan Sun,

1:29:33

host of Close All tabs from KQEDD,

1:29:35

where every week we reveal how the

1:29:38

online world collides with everyday life. You

1:29:40

don't know what's true or not, because

1:29:42

you don't know if AI was involved

1:29:44

in it. So my first reaction was,

1:29:47

ha ha ha, this is so funny.

1:29:49

And my next reaction was, wait a

1:29:51

minute, I'm a minute. is this

1:29:53

is this I think we will

1:29:56

we will see

1:29:58

it to president maybe maybe

1:30:00

within our lifetimes.

1:30:02

You can find Close All

1:30:05

tabs you listen to

1:30:07

podcasts. podcasts.