0:00

Ever wonder what's on a tech startup's Mac?

0:02

Multiple apps and endless presentations

0:05

to make the startup stand out. Mail

0:08

to connect with big name investors. And

0:10

Keynote to craft the big pitch to lock

0:12

them in. Adobe Acrobat to

0:14

sign contracts that secure development funding.

0:17

Xcode to build the app that solves a problem.

0:20

And Todoist to keep the team on track

0:22

for delivery. All from one

0:24

powerful machine. Launch

0:27

a startup on Mac. Support

0:30

for this show comes from Kraken. Crypto

0:32

is like the financial system, but different.

0:35

It doesn't care where you come from, what you look like,

0:37

your credit score, or your outrageous food delivery

0:40

habits. Crypto is finance for everyone,

0:43

everywhere, all the time. Kraken,

0:46

see what crypto can be, not investment

0:48

advice. Crypto trading involves risk of loss.

0:50

Cryptocurrency services are provided to US and

0:52

US territory customers by Payword Ventures Incorporated.

0:55

View PBI's disclosures at kraken.com

0:57

slash legal slash disclosures.

0:59

This is Recode

1:01

Media, Peter Kofka. That is me. This

1:03

is not the episode we were gonna bring you this week.

1:06

We taped that one last week. It was a big look back

1:08

at the year in AI. Kevin Roos from the New York

1:10

Times really helped us sort of walk through the

1:12

hype cycle, the reality. It was great. And

1:15

now we can't use it because of the

1:17

events of the last week. And so Kevin has graciously

1:19

agreed to come back again. He's on no sleep. He may

1:21

collapse during the podcast. That

1:23

could be a good episode. Welcome back, Kevin. That'd

1:26

be great content. Yeah, if I just start snoring

1:29

during the podcast,

1:29

just wake me up, okay? It's

1:32

a good clip at least. We're recording this at 1.34 on

1:34

Monday, November 20. If

1:36

things could change by the time you hear this. But Kevin,

1:39

let's do some really basic stuff. Who is the CEO

1:41

of OpenAI right now?

1:43

As of this moment, and as

1:46

you said, things can change by the minute, it

1:49

seems like. So as of this moment, the CEO

1:52

of OpenAI is Emmett Shear. He

1:55

is the former CEO

1:57

of Twitch and he was appointed.

1:59

last night by the board

2:02

of OpenAI on an intern basis. And

2:07

prior to Friday afternoon, if you just

2:09

turned off your computers and didn't check your phones Friday

2:12

afternoon, you might have thought the CEO of OpenAI

2:14

was Sam Altman. Where is Sam Altman

2:17

now? Well, I

2:19

don't know physically where he is in the world.

2:22

I assume he's at home in San Francisco,

2:25

but spiritually and sort

2:27

of organizationally, he is now

2:29

a, at least purportedly

2:32

an employee of Microsoft. He and

2:35

Greg Brockman, the president and co-founder

2:37

of OpenAI are

2:40

leading a new advanced

2:43

AI research team at Microsoft. Again,

2:46

this is all as of announcements

2:49

made late last night in sort of the fog

2:51

of war, but it appears that he

2:53

and Greg and hundreds

2:56

of other OpenAI employees

2:58

are prepared to decamp to

3:00

Microsoft. So to back up before

3:03

we go any further, last week OpenAI

3:05

was the breakout startup of the AI

3:07

boom. It was worth a theoretical $86 billion. Now

3:12

Microsoft has essentially aqua hired most

3:14

of the talent from there and it's now a unit of

3:17

Microsoft or there's a version of OpenAI that's now

3:19

a version, a unit of Microsoft.

3:21

We still don't know the exact details because again,

3:23

all this is shifting so quickly, but

3:26

what we know is that Microsoft

3:29

CEO Satya Nadella announced on

3:32

Sunday night that Sam

3:35

and Greg Brockman were going to be

3:37

the heads of this

3:40

new AI research

3:42

team, essentially kind of a mini

3:44

OpenAI, inside Microsoft

3:47

and that they had, they

3:51

were doing what they could to bring

3:53

over as many OpenAI employees

3:56

as wanted to come and the

3:58

sort of plan. is that these

4:01

people, unless by

4:03

some 11th hour miracle, Sam Altman

4:05

and Greg Brockman are reinstalled

4:08

at OpenAI, and unless the board of OpenAI

4:11

resigns, the plan as

4:13

of right now is for those people to all go

4:15

work for Microsoft. So this is an amazing

4:18

story in tech. It's the biggest story

4:20

of many, many months. You had to spend the whole

4:22

weekend reporting on it. Again, thank you for coming.

4:25

So we're gonna dig into how all this happened, what it means.

4:28

But if you're just a regular person who's used

4:30

Chat GPT, that's the product OpenAI

4:32

makes, and played with it,

4:34

and maybe they did a homework assignment for you, or

4:37

that's kind of your interest in this, why

4:39

should you care about what happens to

4:41

OpenAI and the fact that OpenAI now

4:43

seems to basically move to Microsoft

4:46

in the middle of the weekend? Well, it matters because

4:48

OpenAI is a massively

4:51

important company in Silicon

4:53

Valley. Not only do they

4:56

make Chat GPT, which

4:58

is the most popular AI product

5:00

on the market, it's got 100 million weekly

5:03

users, but it's also the

5:05

company that's sort of at the front edge,

5:08

the leading edge of the AI boom.

5:10

They are the people

5:13

with the most advanced models, the

5:15

biggest sort of bench of top

5:18

AI talent, and they have been

5:20

kind of setting the agenda for the entire

5:22

tech world over the past year. And

5:25

so it is really a stunning turn

5:27

of events that Sam Altman, who

5:31

has been leading this company through a period of intense

5:33

growth, was pushed

5:35

out by the board in this dramatic, sort

5:38

of mysterious coup led

5:40

by his chief scientist,

5:43

and a number of board members who are also

5:45

affiliated with this movement-effective

5:48

altruism that thinks that AI could take

5:50

over the world and kill us all. It is

5:52

just an insane story and so

5:55

dramatic and cinematic and it's

5:57

almost certainly the most unusual.

6:00

story that I've ever covered in my career

6:02

as a tech reporter. And I want to get into all those

6:04

details, but I just want to channel one of my coworkers

6:07

in Slack today who's not covering this stuff

6:09

but cares about the world. He says, this just

6:11

seems like the same old, same old. There was a company,

6:13

a big important startup. Now

6:15

Microsoft owns a big important startup. It's

6:18

just the same old, same old. That's

6:20

a jaded Monday response, but also I

6:22

get it. Why should you care who

6:25

owns this stuff? Well I think

6:27

it's just intrinsically interesting because

6:29

of all the drama and the back

6:31

and forth. I mean this is not how

6:33

tech business as usual gets

6:36

conducted over a weekend, before

6:38

a holiday, in this kind

6:40

of cloak and dagger way. But

6:43

you've probably seen the social network and

6:45

that looks like guys screwing each other over

6:47

for money and it's a sad story. Maybe you've heard

6:49

about Twitter firing at CEOs one

6:51

after another. It seems like

6:54

if you want to be really reductive about

6:56

it, this is just a bunch of rich white guys messing

6:58

around with each other again. I

7:01

mean if that's how you want to think about it, I

7:03

don't want to dissuade anyone. This is my straw man.

7:05

I'm trying to get you to do it. Right. I

7:07

mean no one has to care about this story. I'm just saying why I care

7:09

about this story and why I think it's actually quite

7:12

important. I also think the

7:14

stakes here are much different than

7:16

the traditional kind of business

7:19

dispute. These are not people

7:21

squabbling over money or control

7:24

of some company. This

7:27

is really a fight between two visions

7:30

of the future. In one of them,

7:32

AI is an exciting

7:35

product and an opportunity to build

7:38

new apps and new systems

7:41

to transform the global economy

7:43

and to sort of lead us into a glorious utopia.

7:46

In the other one, it is a menace

7:49

and a kind of looming

7:51

threat to humanity that is getting

7:54

ever more powerful and that could someday wipe

7:56

us out. I mean it sounds like the plot

7:58

of a science fiction movie. But

8:00

it is really happening. These are really

8:02

the stakes that the people involved think are

8:05

involved here. So for that

8:07

reason, I think it's of interest to a lot of people. So

8:09

let's talk about that fight because

8:11

I think some of this was, if you cared to

8:13

look, was evident but most people weren't paying attention

8:16

to this stuff. When OpenAI

8:18

put out its press release on Friday afternoon,

8:21

it seemed like this might have been a good old-fashioned

8:23

business dispute. They said, Sam

8:25

Altman's leaving and then they normally

8:28

in a press release like that would say he's going

8:30

to pursue other interests, spend more time with his

8:32

family, he's going to stay on as a consultant.

8:34

But they made it very clear that they were firing him and

8:36

they said they were firing him because of a lack of candor.

8:39

So there was a brief period where my phone was blowing

8:42

up with rumors about what Sam Altman

8:44

had done to get himself fired

8:46

by the OpenAI board because it had to be something

8:49

really bad to get fired as

8:51

the head of... You're the face of AI,

8:53

you're the face and founder, you're

8:55

the guy running this $96 billion company,

8:58

you must have done something terrible. But

9:01

it looks like, we haven't heard any evidence that that's

9:03

actually the case and what looks like there's actually this sort

9:05

of theological schism you were just talking

9:07

about. And so you're hearing terms like effective

9:10

altruism, doomerism. Can

9:13

you just get a little deeper into

9:15

how there were two different camps at

9:17

this same company and how they split? Yeah,

9:20

so it's a long and very convoluted

9:22

story and it would take maybe an hour to

9:25

explain the whole thing but I'll just do the... And a flowchart,

9:27

flowchart tend to help. But basically

9:30

here's the deal, there is this movement called

9:32

effective altruism and effective altruism if

9:34

you've heard of it, you've probably heard of it during the Sam

9:36

Bankman freed FTX scandal.

9:39

It's all coming together, it's just the Sam and effective altruism

9:41

means something bad. Yeah, so this is a group

9:43

of people that started

9:46

maybe a decade

9:48

ago in earnest, although the ideas

9:50

are somewhat older than that. They're basically

9:52

trying to use rationality

9:55

and logic to determine how to do the

9:57

most good in the world. And several...

10:00

Several years ago, a lot of people in this community

10:02

got worried about AI risk,

10:04

and specifically what they call X

10:06

risk, which is the risk that AI

10:08

could cause us to go extinct

10:11

as a species. And

10:14

so for various reasons, they decided

10:16

this was a top priority for them. And so

10:18

a lot of effective altruists went

10:20

to work in the field of AI

10:23

and AI safety, especially at

10:25

companies like OpenAI. A

10:27

lot of their early researchers

10:29

were either affiliated with effective altruism

10:31

or sort of sympathetic to their views.

10:35

And these people

10:37

have sort of fallen out of favor with

10:39

OpenAI in more recent months,

10:42

in part because they are seen

10:45

as being overly negative, overly

10:47

pessimistic. Sometimes you hear people

10:49

call them doomers. And so

10:52

there is another crowd inside of OpenAI

10:55

that is more excited about AI

10:57

and less worried about existential risk

11:00

and more commercial in how they're pursuing

11:03

it. They want to build AI into a profitable

11:05

business and build it into popular products.

11:08

So those are the kind of two warring camps

11:10

that have been inside OpenAI

11:12

this whole time and really came to a head last

11:15

week. But Sam Altman has been doing this tour where

11:17

he says, I'm building this new technology. It's really great.

11:20

It could be really dangerous. Is he both

11:22

an enthusiast and a doomer? It seems like

11:25

he's pro enthusiast. He has

11:27

been trying to walk a very careful

11:29

line. So funnily

11:31

enough, you are not the only podcaster who

11:33

had an episode become suddenly

11:36

obsolete last week. My

11:38

co-host and I on Hard Fork

11:40

Casey Newton actually interviewed Sam Altman

11:43

two days before he was fired as

11:46

the CEO. And

11:48

he had no idea any of this was about to happen to him.

11:50

We obviously had no idea. So we had this long

11:53

conversation. But one of the things we talked about

11:55

was this question of whether he is more

11:58

of a doomer or more of a – what they would call

12:00

an accelerationist, someone who thinks that

12:02

AI is good and should move faster

12:05

and we should kind of get out of the way. And he

12:07

said basically he thinks AI is

12:09

going to be amazing, but he also

12:11

thinks that if we don't get it right, it

12:14

could be sort of disastrous. And

12:16

so he's sort of trying to play both

12:18

sides of this debate and really

12:20

listen to and express the concerns of

12:22

both sides. And the other thing that is confusing

12:25

and confounding to a lot of people, including me

12:28

is that OpenAI is a nonprofit

12:31

that also has a for-profit arm.

12:33

So when we talk about this thing theoretically

12:35

being worth $96 billion, that's like a subsidiary of

12:39

the nonprofit org and the nonprofit

12:41

board is the one that fired Sam Altman. Yeah,

12:44

it's a very confusing and convoluted

12:47

structure. But basically, you know, OpenAI

12:49

was started in 2015 as a nonprofit. A

12:53

couple of years after that in 2019, they

12:55

decide, hey, we're building these AI models.

12:58

They're expensive. We need a lot of hardware. We're

13:00

going to need to bring on a lot of investments from

13:03

other companies. They started

13:05

a for-profit subsidiary, technically

13:07

it's a capped-profit subsidiary. And

13:10

that is sort of what people think of as OpenAI.

13:13

But there is still this nonprofit board

13:15

that is in charge of the

13:18

for-profit subsidiary and

13:20

gets to make decisions like whether or not to

13:23

fire the CEO. So now you have

13:25

this structure that is very unusual

13:27

in the tech world. Usually the board is

13:30

composed of VCs,

13:32

investors, early

13:34

employees, top executives, things like that.

13:37

And they don't generally do much. In theory,

13:39

the one thing they can do is fire the CEO. And

13:42

in theory, what they're supposed to do is look out for the

13:44

good of the company and its investors. The

13:47

idea is they're supposed to sort of help the company be

13:49

better and more profitable. That's not the charge here.

13:51

Exactly. So at most companies, the

13:54

board has a fiduciary duty to

13:56

shareholders. At OpenAI,

13:58

since they have this weird nonprofit

14:00

structure, that board, the nonprofit

14:03

board that controls the company, is

14:05

actually not beholden to shareholders at all.

14:07

And in fact, many of the people on the board don't

14:09

own any shares. And they set it

14:12

up that way on purpose, so

14:14

that in the event that the

14:17

sort of humanitarian mission of

14:19

OpenAI to build safe and

14:21

beneficial AGI, wherever

14:24

to come in conflict with more commercial

14:26

priorities, the nonprofit

14:28

would essentially be able to shut it down or at

14:31

least make some big changes. And so there

14:33

is an argument. I mean, most people in my Twitter

14:35

feed are very pro Sam Altman. I think that's

14:37

not surprising. They may actually like him. He's also very

14:40

powerful, very rich, very well connected,

14:42

running this powerful startup. He's the kind of guy,

14:45

the valley rallies behind. But there are folks

14:47

saying, hey, if the nonprofit

14:49

board thinks that Sam Altman is

14:51

behaving irresponsibly and against

14:53

what they're trying to get done, which is responsible

14:56

safe AI, maybe they made the

14:59

right call. Can you make

15:02

a case on behalf of the board along those

15:04

lines? I mean, I think

15:06

if we're trying to steel man the board's

15:09

position, you could say that, yeah,

15:11

this is things working as they

15:13

were designed. The nonprofit

15:16

board got worried about

15:18

the direction of the company, and they made

15:20

a move and took out the CEO.

15:23

I think the thing that's hard to

15:26

reconcile there is two things actually.

15:28

One is that they still haven't explained what the initial

15:31

dispute was over. There was clearly

15:33

something that convinced them that

15:36

this was an urgent threat to

15:38

the humanitarian mission of open AI,

15:40

but they have not communicated what that was

15:43

at all. And so people are just kind of scratching their

15:45

heads. And the second thing is

15:47

that immediately after they made this decision,

15:50

they appeared to start trying

15:52

to backtrack, start trying to think

15:56

about bringing Sam Altman back.

16:00

that looked like that was gonna happen. They

16:02

have not been consistent in their messaging

16:05

at all about this decision, and

16:07

so that's part of what is making people so

16:09

frustrated with them. Including, and

16:11

one of the people saying, oh, I regret this, I'm

16:13

gonna butcher his name, but it's Ilya Skitsavar?

16:16

Sutsavar, yes. Ilya Skitsavar is the

16:19

chief scientist of OpenAI, and

16:22

was basically the driving

16:25

force behind the decision to

16:27

fire Sam Altman, according to

16:29

our reporting and the reporting of others.

16:32

He was the person who sort of brought these

16:34

concerns to the board, and

16:36

now he is backtracking and saying,

16:39

I made a mistake and I regret doing

16:41

that. He put out a tweet

16:43

this morning, said, yeah, I never meant for this to happen. I

16:45

shouldn't have signed that as if it was just sort of a document

16:48

put in front of him. He's also signed

16:50

a document saying, we want Sam Altman back

16:52

or we're gonna quit. So it's

16:54

very confusing. One of the takeaways,

16:57

again, this can change in

16:59

my Twitter feed, is, hey,

17:02

you know what? This kind of shows you that

17:04

the business model that everyone complains

17:06

about in the Valley, where investors,

17:09

VCs, and big companies have a lot

17:11

to say about what happens to these important technologies

17:14

and all in the service of capitalism. Maybe

17:16

that's a pretty good model. Maybe it's a better

17:19

model than having a non-profit in

17:21

that, and for this important stuff,

17:23

and that actually, yay

17:25

capitalism, essentially. Ben

17:28

Thompson, the very smart analyst, had a note out

17:30

basically saying that this morning. Do you

17:32

think this is going to, what effect

17:34

do you think this is gonna have on sort of this kind

17:37

of model? I mean, it's a

17:39

very unique model, so maybe it's one-on-one. Will

17:42

we see something like this again? I

17:44

think it'll be a while before

17:46

we see something like this again with

17:49

this kind of outside investment.

17:51

I think if you are Microsoft, and

17:53

you have just spent $13 billion, if

17:57

you've just invested that much money into a company

17:59

with a company. this kind of wacky

18:01

organizational structure, I think you're

18:03

going to think very hard about doing

18:06

something like that again after seeing how

18:09

easily it can become destabilized. And

18:12

so I think a lot of companies are going to think twice

18:14

before investing in something with

18:17

this kind of governance structure, if something

18:19

like this with this kind of governance structure

18:22

even exists in the future, which I'm not sure it will. We'll

18:24

be right back after a word from a sponsor. Credit

18:27

for this show comes from Kraken. Crypto

18:30

is like the financial system, but different.

18:32

It doesn't care where you come from, what

18:35

you look like, your credit score, or your

18:37

outrageous food delivery habits. Crypto

18:40

is finance for everyone, everywhere, all

18:43

the time. Kraken gives you the tools

18:45

to access this new world of finance. Whether

18:48

you're a pro or just starting out, Kraken

18:50

is the simple, powerful way to buy, sell,

18:53

and learn about crypto. Their intuitive

18:55

trading tools grow with you, empowering

18:57

you to make your first or your 100th

19:00

trade in just a few clicks. And

19:03

if you get stuck, their award-winning client

19:05

support team is available 24-7, along with

19:08

a bunch of educational resources to help you

19:10

on your way. Kraken, see

19:12

what crypto can be, not investment

19:14

advice. Crypto trading involves risk of loss.

19:17

Cryptocurrency services are provided to US and US territory

19:19

customers by Payword Ventures Incorporated. You

19:22

can view KVVI's disclosures at kraken.com slash

19:24

legal slash disclosures. Sure,

19:28

email is dead. But

19:30

wait a second. Think about it. Is

19:32

it really? Intuit MailChimp

19:35

is proud to present a free interactive

19:38

exhibition at the Design Museum in London.

19:41

From September 28th to October

19:43

22nd, email is dead. We'll

19:45

take visitors on an engaging, immersive

19:48

journey through the history of one of the most

19:50

iconic communication tools of

19:52

all time. Email has outlived

19:55

cultural movements, it's outlived technological

19:58

hardware, it's outlived so

20:00

many trends, but it's also

20:02

something that everyone uses, a

20:05

shared, universal part of

20:07

life today. And it's about time

20:09

that it gets a proper tribute. Technology

20:12

will always continue to accelerate. Culture

20:15

will always continue to evolve. And

20:17

it's with that in mind that this groundbreaking

20:19

exhibition asks a bigger question. What's

20:22

the future of email? And

20:24

how will email shape our future? Visit

20:28

MailChimp.com slash emailisdead

20:31

to learn more. And

20:34

we're back. Winners and losers, you did

20:36

a column on that. Let's break down a couple of them.

20:39

Again, things can change, but as

20:42

of now, at 1.51 p.m.

20:44

on Monday, November 20th, what's

20:46

the status of OpenAI? Can it be a company

20:49

if those, I'm seeing now numbers 650 of

20:51

its 700 employees say

20:54

they're gonna leave. Is there a company

20:56

if they leave? I mean, I guess

20:59

there is like on paper a company. It's

21:02

not the same company, certainly, if the

21:05

vast majority of employees leave. And

21:07

it's not even clear what the

21:09

board would be the board

21:12

of. I mean, it's got a commercial

21:14

deal with Microsoft that's still in place, right? That's

21:16

a thing, right? Right, but you know, who would keep ChatGPT

21:19

up and running? Who is going to build

21:22

the next generation of these models?

21:25

And who would want to go

21:27

work for them now? So I think there's a real

21:30

existential threat to OpenAI. Yeah,

21:34

I don't imagine things

21:36

are looking good for them if they don't

21:38

manage to bring Sam Altman

21:40

back and stop this employee exodus.

21:43

And assuming that Sam Altman stays

21:45

at Microsoft, and then again, that was in the report, that

21:48

was Satya Nadella saying, I've hired these guys,

21:50

we're making this happen. He's the CEO of Microsoft.

21:52

So it appears to be happening. Again,

21:54

could change. On the one hand, he sunk $13

21:56

billion into this weird company

21:59

and it all kind of blew. apart. On the

22:01

other hand, if you said you could buy open

22:03

AI for $13 billion, by today's standards,

22:09

that would be a deal of a century. Did Microsoft do really

22:11

well here in the end? Absolutely. I mean,

22:13

they are the biggest winner in

22:15

this whole thing. They managed to turn

22:19

what could have been a real threat to them

22:22

into a real opportunity. Now, they not

22:24

only have this

22:27

existing deal with open AI,

22:29

and they can continue to use open AI's

22:32

models to power their products. But

22:34

they also, if the

22:37

status quo holds, and

22:39

they do end up getting a big infusion

22:41

of people from open

22:43

AI on this new team led by

22:45

Sam Altman and Greg Brockman, they

22:47

will effectively have a new

22:50

open AI that they own 100% of, and whose

22:52

IP will belong to

22:55

them, and they will have control

22:58

over this new team in a way that they did not

23:00

have control over open

23:02

AI under the old arrangement. Why would

23:04

Sam Altman agree to that? In theory, he could

23:06

have gone and opened up shop today

23:09

and had tens of billions of dollars in investment

23:11

capital, in the same way that Elon Musk

23:14

had people just over the transom throwing a billion

23:16

dollars here and there. I imagine Sam Altman

23:19

would have the exact same offer available

23:21

to him right now. Yeah, I've been calling

23:23

around on this question over

23:26

the past like 12 hours or so.

23:28

And I would say there are

23:31

a couple things that I've heard. So Sam

23:33

Altman and Greg Brockman, they could have

23:35

gone out, they could have raised a ton of money. Any

23:38

investor in Silicon Valley would have written them a

23:41

giant check to start a new AI company.

23:44

But starting a new AI company has

23:46

a cost associated with it. You

23:48

know, you've got to form the structure

23:51

and write the article of incorporation.

23:53

And you have to do all this

23:55

sort of paperwork that comes with starting a company.

23:58

But you also have to do things like that. like acquire

24:00

a bunch of compute. You have to go out and buy

24:03

a bunch of GPUs from somewhere, and

24:05

that's not easy to do right now. And

24:07

so with Microsoft jumping

24:10

in, they can sort of hit the ground running.

24:13

Microsoft already has all of

24:16

these data centers filled with GPUs

24:19

that they can use. And so it's just going

24:21

to be a much easier place for them to get

24:23

up and running. And then Sam, now we're

24:25

full on speculation, but one Sam Altman

24:27

famously in tech did not own

24:29

any parts of open AI. Maybe there's

24:32

indirect ownership. I imagine if

24:34

he's going to go to work for Microsoft, he's not going to do

24:36

that out of altruism. He's going to say, actually,

24:38

I would like a significant piece of whatever

24:40

this $9,600 billion, whatever, whatever

24:43

we think this is going to be worth. I'd like I'd like to own

24:45

part of that. Do you assume that Microsoft

24:47

will have to write him some version

24:50

of a very big check to make this happen? I

24:52

don't know. I mean, that's that remains

24:54

to be seen. I can

24:57

speculate that I don't think money

24:59

is all that important to Sam Altman

25:01

at this point in his career. I mean, he's

25:03

already fabulously wealthy. He

25:06

was running open

25:08

AI with no real stake

25:10

in the company, and nothing financial

25:14

to really gain from its success.

25:16

So maybe he doesn't care about

25:18

money, even if he is working for Microsoft,

25:20

or maybe this is sort of his chance

25:23

to, I don't know, get a bigger

25:25

stake in something that he's been working really hard

25:27

on. I don't know the answer to that, but we'll see. Back

25:30

to the bigger picture takeaway, the

25:32

argument that AI is this powerful

25:35

technology that could also doom us has

25:37

been floating out for a while. It's got

25:40

skeptics sort of across the spectrum. They're

25:42

like, there's sort of leftish leaning people

25:44

who say, oh, when you make that argument, you're actually saying

25:46

that AI is more powerful than it is. There's people like Mark

25:49

Andreessen, who say this is doomerism. Right

25:51

now, that argument seems to be associated

25:53

with real corporate ineptitude.

25:57

Again, the story may evolve

25:59

over time, but right now, Now our understanding is the

26:01

board really sort of handled this the

26:03

worst possible way. Does

26:05

this make it harder for anyone to sort of realistically

26:08

argue that AI really is a threat

26:10

that we need to be worried about? It sounds like

26:13

these are sort of like the weird hippies

26:16

complaining about some sort of

26:18

thing that never came to pass and they're stuck in the 60s. These

26:21

guys seem out of touch and clueless. Is

26:23

this going to set that argument back? It could.

26:27

This is all speculation since all this is unfolding

26:29

so quickly. I have talked to people

26:32

who are sort of in the effective altruist,

26:34

what you could call the doomer

26:36

camp. And they're worried

26:39

that this will sort of set back the movement because

26:41

now when people think of effective

26:43

altruism, they might think, oh,

26:45

that's the sort of ideology

26:48

that took down open AI or

26:50

the... And FTX. And FTX, right. So

26:54

I think there is an argument. But I also think

26:56

there's an argument that this was actually a success

26:58

for the movement. That in a very

27:01

real way, this is a movement that has been concerned

27:04

that AI was moving too quickly and

27:06

they wanted a pause. They wanted for

27:09

there to be some interruption to

27:11

the acceleration. And

27:13

this is a pause. It's going

27:16

to be months

27:18

before this new... They've

27:21

thrown their bodies on the gears of the machinery.

27:23

Yeah. And whether you think it was effective

27:25

or not, they certainly made

27:28

an impact and they were

27:30

able to make a strong statement in

27:32

favor of caution and safety

27:35

around AI. One last, and I'm going to

27:37

let you go and you can nap or report or whatever you want to do,

27:39

do both things at the same time. I originally had you on

27:41

because we're basically at the one year anniversary

27:43

of open AI rolling out chat

27:45

GPT, became an immediate sensation. There

27:48

was an enormous amount of consumer popular

27:51

interest in AI for a bit that

27:53

has faded. Pretty considerably. And

27:56

there's a little bit, there had been some sourness about

27:58

sort of maybe what AI means for the vast... If

28:00

we step out of all that, we're now a year

28:03

in the sort of AI being mainstream.

28:06

What are you most excited about as someone who covers

28:08

this stuff all the time? Aside from

28:11

this soap opera on steroids, what

28:13

are you most excited about when you think about AI? Oh,

28:15

man. It's hard to

28:17

pull my brain out of the soap opera. It's

28:20

been an all weekend. I mean, you

28:23

know, I think I would say I'm really

28:26

interested in just how AI

28:28

is working its

28:33

way into normal people's lives.

28:36

I'll have talked to doctors

28:39

and lawyers and teachers and musicians

28:41

and filmmakers and all kinds of people who are

28:44

using this stuff, grappling with the

28:46

ethics of using this stuff. And

28:49

it's just been really fun to

28:51

kind of see how this technology that

28:54

kind of arrived unannounced

28:56

and with very little fanfare has

28:59

just totally upended

29:01

some portion of the world.

29:05

And yeah, it's actually like something that I was

29:07

excited to ask Sam

29:09

Altman about before all of this went down.

29:12

What was his answer, by the way? If I can plug

29:14

my podcast. Because

29:16

we're going to be also releasing the

29:19

interview that we did with Sam Altman last

29:21

week on our show Hard Fork today.

29:24

So you can go listen to that if you want. But

29:27

he was just sort of like, look, we

29:29

didn't know how big this was going to be, but

29:31

it's been really gratifying to see all of the

29:34

interesting and useful and heartwarming

29:37

and weird ways that people are using

29:40

this. And people argue about

29:42

what he said was he doesn't really like

29:44

this sort of argument about AGI

29:47

and what counts as being superhuman

29:49

intelligence. His basic message was, look,

29:51

we built a thing and people

29:53

find it useful. And that's all that matters. Kevin

29:56

Roos, it is 2 0 1. We're going to end the

29:58

conversation here if this thing Everything is obviated

30:00

by the time we publish. We tried our best. We

30:03

did it twice, hopefully the second time

30:05

around works. Kevin Reuss of the New York Times

30:08

writes the shift column there, does great work there, also

30:10

does, as he said, Hard Fork, the great podcast

30:12

with him and Casey Newton. Thank you, Kevin. Thanks,

30:15

and I'm not coming back a third time. See ya. Thanks

30:19

again to Travis and Shalani and Julie Myers

30:21

as well. Thanks to our sponsors and

30:23

thanks to our listeners. This is Recode

30:26

Media. We'll see you soon. Support

30:30

for this show comes from Kraken. Crypto

30:32

is like finance, but different.

30:35

It doesn't care when you invest, trade,

30:37

or save. Do it on weekends or

30:40

at 5 a.m. or on vacation

30:42

at 5 a.m. Crypto is finance

30:44

for everyone, everywhere, all

30:46

the time. Kraken, see what crypto

30:49

can be, not investment advice. Crypto

30:51

trading involves risk of loss. Cryptocurrency

30:53

services are provided to U.S. and U.S. territory

30:55

customers by Payword Ventures, Inc. View

30:57

PBI's disclosures at kraken.com. As

31:02

a marketer, you're speaking to a vast

31:05

audience. Some people need to be converted

31:07

into customers, some need to be reunited

31:10

with their carts, and others have just

31:12

made a purchase. But when you fail

31:14

to segment your audience and personalize

31:16

your messaging, you can get what's called

31:19

a cluster, one big cluster

31:22

of customers who may seem alike but

31:24

actually all have different behaviors. So

31:27

how do you turn those customers back

31:29

into customers? With Intuit

31:32

MailChimp, you can use personalization

31:34

tools that segment customers into groups,

31:37

break them up into like-minded target audiences,

31:39

and send them personalized marketing. Intuit

31:42

MailChimp, the number one email marketing

31:45

and automations brand. Based

31:47

on competitor brands publicly available data

31:50

on worldwide numbers of customers in 2021 and 2022.

31:54

Availability of features and functionality vary

31:56

by plan, which are subject to change. you