HomeVideos

Live from Bloomberg Tech in San Francisco: Android Hardware, Mozilla's Open Source AI & Venture...

Now Playing

Live from Bloomberg Tech in San Francisco: Android Hardware, Mozilla's Open Source AI & Venture...

Transcript

1771 segments

0:00

[music]

0:02

Bloomberg Audio Studios, podcasts,

0:05

radio, news.

0:08

This is Bloomberg Business Week Daily,

0:11

reporting from the magazine that helps

0:13

global leaders stay ahead with insight

0:16

on the people, companies, and trends

0:18

shaping today's complex [music] economy.

0:20

Plus, global business, finance, and tech

0:23

news as it happens. The Bloomberg

0:25

Business Week Daily podcast [music] with

0:27

Carol Masser and Tim Stenbec on

0:30

Bloomberg Radio.

0:32

>> Carol Master, Tim Stenbec. Uh this is

0:34

this annual gathering of leading CEOs,

0:37

innovators, um everybody who is

0:39

harnessing technology to really change

0:41

the world around us. Safe to say once

0:43

again, we're talking a lot about

0:44

artificial intelligence.

0:45

>> Yeah, we've heard from the executives,

0:46

the venture capitalists, the founders at

0:48

the forefront of the next era in

0:49

technology, AI, wearables, robotics, the

0:52

blockchain, and more. Someone who knows

0:54

a lot about this space is our own Brad

0:56

Stone. He's sitting down next to me.

0:57

We're going to get to him in just a

0:58

moment. Um, we're also going to talk

1:00

with our own Ed Lello. He's fresh off a

1:02

conversation with the co-founder and

1:03

executive chairman of Anderal

1:04

Industries. We're talking about Trey

1:06

Stevens. So, we'll talk about that.

1:08

Also, a really fascinating conversation

1:10

um with Cerebrus Systems last night. So,

1:14

Ed's going to also dig into that as

1:15

well. So, um that's coming your way.

1:17

>> Yeah. Joining us too uh a little later

1:19

is Katie Han, the former federal

1:20

prosecutor. She was the first female

1:22

general partner over at Injury Suborit.

1:24

She's going to join us. She's got Han

1:26

Ventures. Just raised a billion dollars

1:28

for a new venture fund known for crypto.

1:30

Yes.

1:30

>> But branching out of crypto a little

1:32

bit, but really also just staying in her

1:33

lane.

1:34

>> Maybe looking for the next reverse.

1:36

>> Yeah, exactly.

1:37

>> Or something like that.

1:37

>> If you can find it, let me know.

1:39

>> Exactly. Hey, let's get right to Brad

1:40

Stone. He's the editor of Bloomberg

1:41

Business Week magazine. He's also the

1:43

author of many, many books, including

1:45

two on Amazon, the Everything Store and

1:47

Amazon Unbound. He also wrote a book

1:50

about the upstarts, which I want to ask

1:51

him about today's upstarts. Anyway,

1:53

let's get right to it. Good to have you

1:55

here.

1:55

>> How many of these have we done?

1:57

[laughter]

1:57

>> A lot. And they're so excited. There's

2:00

so much coming at us. Um, you walk in

2:02

here and what's interesting for you?

2:03

>> Well, I mean, today is really in large

2:05

part a conversation about AI. And every

2:07

time you talk about AI, you have this

2:09

tension between sugar coating and

2:12

cander, optimism and concern. And that

2:16

is no different. I mean, we heard from

2:17

Yosua Benjio, a professor, University of

2:20

Montreal, a godfather of AI, saying we

2:22

cannot trust the big tech companies to

2:24

do the right thing. And then Fe Lee, a

2:26

godmother of AI, saying that she

2:28

believes that all this friction is kind

2:30

of overblown, that it sucks all the

2:32

oxygen out of the room and you have

2:34

these polar opposites that are

2:36

unrealistic. And then Dan Schulman, CEO

2:38

of Verizon, saying, "Yeah, a lot of the

2:40

customer service employees are gone." I

2:43

thought he spoke with a lot of cander

2:44

today.

2:45

>> Yeah. Where where are we in a cycle

2:48

right now, Brad? Because you've, as

2:49

Carol mentioned, you have the book about

2:50

the upstarts. You've written about

2:52

Amazon. You've written about tech

2:53

companies and a completely different

2:55

part and era of Silicon Valley. Are we

2:57

in a cycle right now that's any

2:59

different than others?

3:00

>> Tim, you I think you know the answer to

3:01

that one. We are riding the wave. We are

3:04

surfing it. We are I It feels like at

3:07

the peak of a massive hype cycle.

3:09

>> Does it feel like this time is

3:11

different? Those dreaded words that I'm

3:13

never allowed to say, but I always do. I

3:14

I mean I would compare it to uh the

3:17

internet boom of the late 90s in which

3:19

there's a lot of hype, a lot of froth.

3:21

Not all the companies are real and are

3:23

going to make it. But that you know that

3:25

the underlying change is so fundamental

3:28

and and AI I think you know it it it

3:31

changes everything. It's already changed

3:33

the way we work, the way our company

3:35

works. I think it's happening across the

3:37

board and we're just at the beginning of

3:38

realizing the promise

3:40

>> you know but I also think about you know

3:42

the intersection of technology and money

3:45

and people spending a lot chasing to be

3:48

the first the best in this this race but

3:50

I wonder how much of that clouds the

3:52

judgment of whether or not that race

3:55

makes sense in the end

3:57

>> right and we're so we're talking about

3:59

perceptions and valuations

4:01

>> but um look I mean we had Daniela Amod

4:05

day, the the co-founder of Enthropic,

4:07

you know, she gets on stage and says,

4:09

you know, this is this buildout's

4:11

costing a lot of money, so we're going

4:12

to go go tap the public markets. But and

4:14

we've already seen though that those

4:16

offerings are generating a lot of

4:17

excitement. The excitement is real, and

4:20

it has paid off for different investors

4:22

at different levels of the ecosystem.

4:24

You know, the big question ends up being

4:26

whether that retail investor at the end

4:28

of the line, how well they do. Brad, I

4:30

also feel like it's changed San

4:32

Francisco. And you know, Carol and I

4:34

come here a couple times a year and we

4:36

came here during the depths of the

4:37

pandemic. A lot of people left the city

4:40

for dead at that time. And I got to tell

4:42

you, I'm I'm in a small part of it right

4:43

now, but anecdotally speaking, there was

4:45

a 20-minute line for like $7 matchas

4:48

this morning. And everybody was waiting

4:50

in line. There are people everywhere.

4:52

There are people spending money. I walk

4:54

out of my hotel room and I see a

4:55

supercar. Like, what is happening to the

4:58

city right now? Well, first of all, are

4:59

we to understand that you yourself

5:01

purchased a $7 matcha?

5:03

>> I'm just doing research. I'm just doing

5:05

research.

5:05

>> Well, look, we in San Francisco,

5:07

>> you could also get coffee there, too.

5:08

>> We are very good at ringing our hands

5:10

about things. And 5 years ago, it was

5:14

homelessness and the desertion of

5:16

downtown and the chaos of the open drug

5:18

markets. And, you know, credit to our

5:21

new mayor and the change in the business

5:24

cycle. There's a lot of excitement and

5:26

energy in the city. But now the concerns

5:28

are different. We worry that with this

5:30

wave of wealth coming from these IPOs

5:33

that it will be impossible for people to

5:35

buy and rent homes. It's already uh

5:38

classically difficult in the city. And

5:40

it's only going to get more difficult.

5:42

If you're an employee at one of the big

5:44

companies in San Francisco that's not in

5:46

tech, you know, the Gap Levis, how do

5:48

you possibly compete and live in the

5:50

city?

5:51

>> That's a really good point. And so,

5:52

look, I mean, I I think that yes, the

5:54

problems are different and there's a lot

5:55

of energy and excitement, but a city

5:58

like San Francisco, it's hard to kind of

5:59

bottle the economic phenomena.

6:02

>> You bring up a really good point. Levis

6:03

is a great example. How does Levis hire

6:06

a computer programmer or an engineer

6:08

>> if if that person could go get a job

6:10

that, you know, pays 3x and gives stock,

6:13

you know, based compensation over at

6:14

Anthropic?

6:15

>> I think it's incredibly challenging. You

6:16

also have an anti-b businessiness

6:18

environment here that occasionally does

6:21

raise its head. And in the primary that

6:23

we just had earlier in the week, there

6:25

was consideration given to an extra tax

6:27

on businesses whose CEO makes a factor

6:30

larger than average employees. Now, that

6:32

ballot did get defeated, but it's

6:34

there's a lot of push and pull around

6:36

what kind of city do we want to be and

6:38

how do we manage the economic

6:40

disparities that are being created by

6:42

this AI boom.

6:43

>> Well, go into politics. I feel like

6:44

you've been having a lot of

6:45

conversations. We think about the the

6:47

governor's race, some interesting

6:49

candidates in that. I mean, what do the

6:51

people of California want going forward

6:53

and who's going to do that for them?

6:55

>> I mean, the funny thing is we voted on

6:56

Tuesday and we do not know what the the

6:59

citizens of California said yet because

7:02

only about half the votes have been

7:03

counted, but so far Javier Bera, Scott

7:06

Hilton are the leaders. You've got Tommy

7:08

Styer still within shouting distance.

7:10

Yeah, it's interesting because some of

7:11

the most extreme viewpoints on tech seem

7:14

like they're not advancing. I me

7:16

mentioned the ballot initiative that's

7:18

failed. You know, Styer was really the

7:20

most aggressive about regulating tech

7:22

and AI. He doesn't seem to have made the

7:24

runoff. Of course, we don't know. Um, so

7:27

but look, I mean, for some reason, we

7:28

count the ballot slowly here, but I I

7:32

will be interviewing um uh Scott Weiner,

7:35

Senator, State Senator Scott Weiner, who

7:37

appears to be the front runner for Nancy

7:39

Pelosy's job. He will instantly become

7:41

the most interesting and maybe important

7:43

regulator of tech and AI in the nation's

7:46

capital. If he wins the election in the

7:48

fall,

7:48

>> I have to say, and that is one of those

7:49

things that has come up often with the

7:51

naysayers or those who are concerned. um

7:53

whether it's Joshua Benjio, I mean just

7:55

concerned about you know having people

7:58

in government, the United States

8:00

government but also globally to make

8:02

sure that the regulations and the

8:04

oversight is there. What do you want to

8:06

hear from him

8:07

>> from Senator Weiner? Yeah.

8:08

>> Yeah. Well, um I think two things. One,

8:10

he has been uh an important state

8:13

legislator. He's proposed these two

8:15

bills about regulating AI. The one that

8:17

passed um was the watered down version.

8:20

I want to know that if he does get

8:21

elected, what kind of AI regulation does

8:23

he bring to the nation's capital? But

8:25

the second thing is that some of the

8:27

ideas we hear right now for regulating

8:28

AI come from Senator Bernie Sanders and

8:31

they're pretty extreme. A data center

8:33

moratorum and state ownership, 50%

8:36

ownership of the AI companies.

8:37

Obviously, Silicon Valley here does not

8:40

like those ideas. So, how does he work

8:42

with his colleague,

8:43

>> right?

8:44

>> We have uh five or you know 8% state

8:46

ownership of Intel right now, right?

8:48

>> We never thought that would happen.

8:49

Well, Senator Sanders is talking about

8:51

50% ownership at the same time as a data

8:54

center moratorum would really slow down

8:56

the AI revolution and crush valuation.

8:58

>> So, so on this political question, what

9:01

what's the narrative right now around

9:03

finding employees outside of Silicon

9:05

Valley to do the work

9:07

>> that Silicon Valley is so known for? And

9:09

this idea of taking like Elon Musk

9:11

taking his company's headquarters to

9:14

Texas or people relocating to Miami. Is

9:17

Silicon Valley still where the talent

9:20

is?

9:21

>> I mean, absolutely. I think a couple

9:22

years ago, we heard about the migration

9:24

away from San Francisco or Silicon

9:26

Valley because of an anti- tech or

9:30

anti-billionaire environment here.

9:32

There's still a looming question about

9:33

the billionaires tax which will go in

9:35

front of voters in the fall. But no,

9:37

this is the concentration of innovation

9:40

and ingenuity is still here. But you

9:42

mentioned the narrative um and I I think

9:45

that the narrative particularly around

9:46

regulating AI has changed. We recently

9:49

saw President Trump introduced an

9:51

executive order on AI that called for a

9:53

30-day moratorum on or 30-day review

9:56

period on reviewing AI models obviously

9:59

because of mythos and the concerns about

10:02

v vulnerability. Now, that was down from

10:03

a 90-day review period, but for the

10:05

first time, the president has gotten

10:07

farther ahead than a lot of tech

10:09

companies and legislators hear about the

10:12

need to be to go a little bit more

10:14

slowly and with introspection about

10:16

rolling out these powerful models.

10:18

>> Well, speaking of them being powerful

10:19

models, I know Mary Daly, president of

10:21

the San Francisco Fed, was up here uh

10:23

talking with Caroline Hyde. She said

10:25

there's still no clear evidence in the

10:27

economic data productivity gains from

10:29

AI. She remains bullish on the

10:30

technology, sees possibilities for early

10:32

rewards. But I think about the irony or

10:34

the harsh reality perhaps that if this

10:37

does make us all product much more

10:39

productive, takes away jobs like what

10:41

does it mean for Silicon Valley or the

10:44

rest of the country,

10:45

>> right? Well, we've heard a lot of

10:47

opinions about that. What I thought was

10:49

interesting from Mary Daly is she was

10:50

comparing it to the introduction of

10:53

electricity and how it took a while for

10:55

companies for inventors to completely

10:58

rethink their technologies and business

11:00

models. And that's what you said that we

11:02

haven't yet baked in the power of AI

11:04

fully into our companies. But I mean in

11:07

terms of um yeah what AI does for jobs I

11:10

mean Dan Schulman of Verizon saying uh

11:13

after all the layoffs they've had that

11:15

is essentially coming for all their

11:17

customer service workers to an extent.

11:19

Um we we heard it from Trey Stevens of

11:22

Andreel um uh certainly from from Yosua

11:25

Venio. Yeah I mean I think like there's

11:28

still a a really interesting

11:29

conversation here. how to what extent it

11:32

displaces workers and if so what is the

11:34

responsibility of government government

11:36

to come in and either retrain workers or

11:40

you know help with the livelihood for

11:41

everyone who's displaced.

11:43

>> We're speaking with Brad Stone. He's the

11:44

editor of Bloomberg Business Week. He's

11:46

on set here at Bloomberg Tech here in

11:48

San Francisco, the Bloomberg Tech

11:50

Summit. I want to go back to to our

11:51

interview with Joshua because what what

11:53

I found so fascinating was that, you

11:56

know, he's he's known as one of the

11:57

godfathers of AI and and he he there's a

12:00

lot of stuff that really keeps him up at

12:02

night and he couldn't really tell us for

12:03

certain if the potential positives of

12:06

the technology outweigh the potential

12:08

pitfalls of the technology. And I'm just

12:11

wondering given your experience covering

12:13

this stuff if we've ever seen a moment

12:16

where we're sort of at this precipice

12:17

like this. I mean, I think Mythos from

12:20

Antropic was was, you know, an

12:22

inflection point. The fact that we have

12:24

this technology that can identify

12:28

vulnerabilities in all of our important

12:30

systems and of course remediated. But

12:33

the question is, you know, when that

12:35

technology is broadly distributed, is it

12:38

exploited before it's remediated? And

12:40

what uh what he said was that mythos is

12:44

the beginning that everybody catches up

12:46

to mythos eventually even the Chinese

12:48

models and also I can't remember who out

12:51

somebody else said that that open AI

12:52

will be here soon and other companies

12:54

and so yeah I think we're we've hit a

12:56

point where the the genie is escaping

12:58

the model.

12:59

>> Yeah, Rafie Coran from Mozilla was on

13:01

our program and talking about the way

13:02

that they've used it to identify

13:04

vulnerabilities at at Firefox. Um but at

13:07

the same time uh you know we're mythos

13:11

is going to be old news in a year from

13:12

now when we're talking to you

13:13

>> right and

13:14

>> and everybody else is going to be

13:15

>> and and every commercial model will

13:17

essentially be mythos scale and it will

13:19

be the age of the zero day vulnerability

13:21

where these these openings are exploited

13:24

before companies or governments could

13:26

possibly rectify them and so um you know

13:29

and then it was interesting Daniela

13:30

saying that they will begin to release

13:32

these commercial models before consumer

13:35

models. So obviously the industry is

13:37

starting to think about and of course we

13:38

mentioned the Trump executive order

13:40

regulating these rollouts and trying to

13:42

instill some order around how these

13:44

tools are distributed.

13:46

>> Um you've covered technology for a long

13:48

time. You've seen similar events like

13:50

this from Bloomberg. I mean do you ever

13:53

walk away because it feels like a lot of

13:54

the tech community is always optimistic

13:56

about what's to come. Can you can you

13:58

feel a difference from last year or two

14:01

years ago in terms of tones and and what

14:03

we're hearing from the industry?

14:05

>> I feel like in some years

14:08

>> only in retrospect do I think boy we

14:10

were scrambling a little bit to make

14:12

that interesting.

14:13

>> You know that we that that obviously the

14:15

the issues of the day they loom large

14:17

but in retrospect maybe they weren't as

14:19

significant as we thought and this year

14:21

I I and and it was the same last year as

14:24

well. You know, it feels like people are

14:26

here legitimately because they want to

14:27

learn, because they they feel

14:29

desperately behind, because news and

14:32

events are happening faster than we can

14:34

possibly integrate them into our our

14:37

lives. And I I feel like that's the

14:38

energy here today. People want to hear

14:40

from these speakers and they want to

14:42

understand how they can bring this into

14:43

their own lives.

14:44

>> It's funny. I feel like you and I were

14:46

talking a little bit last night or like

14:47

I feel that way. you know, I've

14:48

scratched the surface in terms of what I

14:50

can do, but I want to understand it more

14:52

>> because yeah, we had this conversation

14:53

at dinner like none of us are vibe

14:55

coding right now and I feel like we're

14:56

just using the LLM in the most sort of

14:59

like elementary way as glorified search

15:02

and I'm like I'm not really there yet,

15:03

>> right? So, I think there's a real hunger

15:05

to learn. I think we could make this a

15:07

three-day event and have bring in people

15:09

to teach teach people how to buy code,

15:12

how to use the latest models. I I don't

15:16

I'm not in charge. I don't I don't. Um,

15:19

but I try, you know, when I can, I try

15:21

to bring it into my work. Are

15:23

>> are Okay.

15:25

Are are your kids vibe coding?

15:27

>> Well, we got a daughter back here, so

15:28

she could answer.

15:29

>> Are you vibe coding?

15:30

>> I don't know what that is.

15:31

>> Okay. Fair enough.

15:33

>> The future is safe. [laughter]

15:34

>> The kids are all right. The kids are all

15:37

right.

15:37

>> Oh boy.

15:38

>> Um, if you could talk to one person for

15:40

an hour right now when it comes to the

15:42

world of AI, who would it be? I mean

15:44

there's so many um you know there

15:46

there's so many great speakers here

15:48

today. So any but anybody

15:50

>> anybody doesn't have to even be here.

15:52

>> I mean that's um yeah that's a real

15:55

tough one. But um Dario Aod and I mean

15:59

they have look how many times in this

16:01

industry does the company that is

16:03

perceived as the secondary player

16:06

suddenly leaprog into the league command

16:09

a higher valuation dominate the

16:11

narrative with its not just its consu

16:13

consumer but it's commercial models and

16:15

I think that's a really interesting

16:17

company to kind of understand he has

16:19

steered that company particularly well

16:21

amid all the chaos of the noise

16:23

>> I will say it's always that moment of

16:25

time like this I wonder okay who of the

16:26

big names that we talk about are not

16:28

going to be here and who are the ones

16:29

that like five or 10

16:31

>> this is a a curse of having an annual

16:33

event we've had Daria we've had Sam

16:35

Alman um you want to try to uh create a

16:38

fresh lineup every year

16:39

>> great stuff thank you so much um really

16:42

appreciate it going all around uh

16:44

Bloomberg Tech here in San Francisco.

16:47

>> Stay with us. More from Bloomberg

16:48

Business Week Daily coming up after

16:50

this. [music]

16:54

>> You're listening to the Bloomberg

16:56

Business Week Daily podcast. Catch

16:58

[music] us live weekday afternoons from

17:00

2:00 to 5:00 Eastern.

17:01

>> Listen on Apple CarPlay and Android

17:03

[music] Auto with the Bloomberg Business

17:05

App or watch us live on YouTube.

17:09

I got to say in prepping for this

17:10

segment [music] I actually found a story

17:12

back in 2019 that Bloomberg reported

17:14

out. They did a deep dive. The story is

17:16

entitled the smartphone revolution was

17:17

the Android revolution. It was by Shira

17:20

Oviday and it got into how the mobile

17:22

operating system called Android drove

17:24

Steve Jobs at Apple crazy. And the

17:27

reporting went on to note that while

17:28

Apple sparked the modern smart

17:30

smartphone revolution, Android Tim was

17:32

the essential ingredient that made the

17:33

devices ubiquitous.

17:34

>> Okay, so fast forward just a few years

17:36

to today. 7 years later, the open source

17:39

mobile platform, it's the world's most

17:40

widely used, depending on the stats that

17:42

you look at, just about 70% of global

17:44

share. Leading the global teams behind

17:46

it, Samir Samat, president of Android

17:48

ecosystem. He joins us live here at

17:50

Bloomberg Tech Tech. Welcome. How are

17:52

you?

17:53

>> Thank you. I'm great. Nice to see both

17:54

of you.

17:54

>> It's good to see you as well. Um, look,

17:56

we think about Android in the context of

17:59

iOS, I think a lot of people, because

18:00

that's that's really the duopoly that

18:04

exists. I think many people would argue.

18:06

How do you position Android in the

18:09

United States versus how you position it

18:11

globally where it is more widely

18:13

adopted? Well, Android is an open-

18:15

source operating system which is really

18:17

great because it allows people to build

18:19

all kinds of different devices. So, of

18:21

course, we know that uh Android is

18:23

popular for phones today where there's

18:24

over, you know, three and a half billion

18:26

Android devices that are used every day.

18:28

It's the most widely used operating

18:29

system in the world. But Android is so

18:30

much more than phones. People use it for

18:32

cars, they use it for televisions, they

18:34

use it for uh for smart watches and all

18:36

kinds of new devices, including some

18:38

smart glasses that we're working on for

18:40

for later this year as well. So, it's

18:42

really become much more than phones and

18:44

it's a whole ecosystem.

18:45

>> Well, you know, it's funny before we got

18:46

started, too. When we think about an

18:48

operating system, how do you think about

18:50

an operating system within an AI world

18:52

that is we're still kind of figuring it

18:54

out, right? We're all finding our way

18:55

here.

18:56

>> Well, it's uh it's changing a lot. You

18:57

know, things are changing all around us.

18:58

And I think the way we think about this

19:00

moment is you we're moving from an

19:02

operating system to what I call an

19:04

intelligent system. Um, one that is able

19:08

to be more predictive, uh, and more

19:10

helpful in the moment for you. So

19:12

instead of, you know, making you the

19:14

micromanager of your device, which is

19:16

the way I think computing's worked for

19:18

40 years, it's you have the idea in your

19:19

head and you use a pointing device or

19:22

taps to take that idea and do all the

19:24

things you need to do to get things

19:25

done. in this new AI era, how can we

19:28

actually make the system much smarter?

19:30

So, you give it your intention, your

19:32

goal, and it can translate that into the

19:34

action that it needs to take to help you

19:37

get more done. Yeah, I want to push on

19:38

that a little bit because I I still

19:39

think when when right now when a lot of

19:42

people think about AI, they they equate

19:43

it with LLMs and they think about LLMs

19:46

not necessarily as, you know, uh,

19:49

clogged code or or or helping us buy

19:52

code, but rather as glorified search

19:54

engines and that actually does search a

19:57

lot better than than we've grown up

19:59

using. H how what is the value

20:01

proposition for an ecosystem in the

20:04

world of AI? because you're explaining

20:05

this idea of us being more productive as

20:08

a result of this technology, but I don't

20:11

even know what that looks like. Is it

20:12

predictive? Does it know what we want

20:14

before we want it? Like, what is it?

20:16

>> Which is gets a little creepy then

20:18

potentially potentially. Potentially.

20:20

Let me give you one simple example that

20:21

I used the new release of Android with

20:24

Gemini for recently on my on my uh Pixel

20:28

device, which is, you know, I had a list

20:29

of folks that are coming to a barbecue

20:31

that I'm holding and next to each person

20:34

I had how many people are coming in

20:35

their family. Are they vegetarian? Are

20:37

they not vegetarian? You know, the

20:39

normal thing you do when you're throwing

20:40

a party, you make this list.

20:41

>> But like where do you put that list?

20:42

Does that go in Google Docs or

20:43

something?

20:43

>> I think I just had it in an email that I

20:45

was writing to myself, you know, like we

20:47

all do sometimes. And I asked Gemini

20:50

working on my Pixel device, can you take

20:52

this list, analyze it, figure out how

20:56

many hamburgers, how many hot dogs, how

20:58

many buns, and then go to Instacart and

21:00

add it all to my shopping cart, and then

21:02

let me check it out. Okay? And you know,

21:05

it took that, figured out all the

21:06

vegetarians, figured out all the things,

21:08

figured out how many people are coming,

21:09

how many kids, how many not, and it did

21:10

it for me. And so that probably saved

21:12

me, I don't know, six, seven minutes.

21:14

You add that up multiple times a day

21:16

throughout all the things that we're

21:18

doing and that's really helpful.

21:20

>> Samir, what do you think in terms of the

21:22

conversations that we're having about

21:23

artificial intelligence? Cuz I do feel

21:25

like you and I were talking about this

21:26

at dinner last night. We're just

21:28

scratching the surface in terms of

21:29

certainly how we use it and how we need

21:31

to push ourselves to better understand

21:33

what this tool can do or eventually do.

21:37

Well, I think the most important thing

21:38

right now is uh for all of us that are

21:40

in the industry or even around the

21:42

industry is to make sure we have on

21:44

hands-on experience. You know, uh one of

21:46

the things we tell our our teams and I

21:48

tell my kids as well is we should engage

21:50

with the technology to understand how it

21:53

can help you and what the limits of it

21:54

are and it's changing so quickly. What I

21:56

tell folks is even if it wasn't able to

21:59

help you with something this month, put

22:02

a note for yourself and check back 3

22:04

months later because it's moving so

22:05

fast. What what I think is important

22:08

about your example is that is that you

22:10

you had the initiative to actually ask

22:13

Gemini to do that. I think that the

22:16

challenge in this world is that a lot of

22:18

people don't know that LLMs like Gemini

22:21

can actually do that sort of thing. So

22:24

they wouldn't even know that with that

22:27

example AI could save them six or seven

22:29

minutes. How do you communicate that to

22:32

to a a a population that has no idea

22:36

what this technology can do? Truly has

22:37

no idea.

22:38

>> Well, I think that's a great question.

22:39

And you know, in the beginning when

22:41

these interfaces came out, you had a

22:44

chat prompt and we didn't know what we

22:47

could type into it. And I think people

22:48

do discover over time by by trying

22:51

things. But most importantly, the

22:53

software itself, this is where the

22:55

operating system can be really helpful.

22:57

You know, you when you talk about an

22:58

intelligence system, it should be able

23:00

to understand from you what your

23:03

intention is, you know, what's my goal?

23:05

What am I trying to get done? Oh, I'm

23:06

planning a party. Okay. Well, did you

23:08

know I can help you with this? You know,

23:10

and to prompt you to do that and to to

23:13

show you how that it can assist you in

23:15

that moment. And that's a different kind

23:17

of user interface than what we've seen.

23:18

>> What I have noticed is with Google

23:20

search, it's it's actually pushing me to

23:23

Gemini within within web- based Google

23:25

search. and it's becoming very natural.

23:28

So, it's sort of like I'm not starting

23:29

at Gemini. I'm starting at Google Search

23:31

and then using it interacting with

23:33

search like an LLM.

23:34

>> Well, I'm glad it's helping you.

23:36

[laughter] That's great. You know, we

23:37

have we have so many different tools

23:39

that that that I think are are there for

23:41

folks. And I I just want to uh give one

23:43

small example of one um called Notebook

23:46

LM. And if you're um if your viewers

23:48

haven't checked out Notebook, people

23:49

love this.

23:50

>> Yes. I mean, Notebook LM is one of my

23:51

favorites. And um you know on the

23:53

Android app for example for Notebook LM

23:55

my son who just finished finals um he

23:58

was like dad can you help me study for

24:00

for the science final um and I said sure

24:04

uh overly confident as I am about this

24:06

[laughter] and then he gave me the

24:07

material and I looked at it and I said I

24:09

haven't reviewed this in years you know

24:11

but tonight I have to do a study session

24:13

with him. I put it all in notebook LM

24:15

and it gave me an audio podcast that I

24:17

listened to on the way home and when I

24:19

got home I looked like super dad, you

24:21

know, because I could answer these

24:22

questions and go back. This is where it

24:24

starts to become helpful and I think

24:26

it's all about in the the purpose that

24:29

the the way in which we use this

24:31

technology in a purposeful, intentful

24:34

way to help make our lives better. And

24:36

that's what we're trying to do. So,

24:37

let's go there because last month when

24:38

you guys announced a bunch of upcoming

24:41

features for your Android 17 operating

24:43

system, you made some comment about

24:45

there's a great concern about AI for AI

24:47

sake and you want to be very purposeful.

24:50

What does that mean?

24:51

>> Well, I I mean I think if you if you

24:53

look at any just go talk to people, the

24:56

word AI has become pretty overloaded and

24:59

I think people are they have mixed

25:02

feelings about it. Let's just put it

25:03

that way. Um, I think what they really

25:05

want is to not hear about the

25:07

technology. What they really want is to

25:09

hear about how is this going to help me.

25:11

Um, I gave the notebook LM example, but

25:14

you know, there there's so many more

25:15

that when you when you talk about how

25:17

this can help you, it it it changes the

25:20

conversation. I'll give you one of my

25:21

favorites from Android. Android has a

25:23

feature called circle to search. Okay?

25:25

And circle to search is really simple.

25:28

You'll hold down the home button and

25:30

then you see a flash and at that point

25:33

you can circle anything on your screen.

25:35

Let's say you you have a celebrity on

25:37

your screen wearing a great outfit.

25:39

Circle it and say shop the look and it

25:42

bring Google will go take what they're

25:44

wearing and figure out what the jacket

25:46

is, what the tie is, what all the

25:48

different pieces and bring back each of

25:50

those to you so you can figure out where

25:52

to buy them and if they're right if

25:54

they're right for you. There's a ton of

25:56

AI behind the scenes there. That's not

25:58

the important thing. The important thing

25:59

is that it's super helpful, packaged in

26:01

a way that's intentful, purposeful for

26:04

the user to help them get something done

26:05

that they couldn't do before.

26:06

>> Samir, where do you think AI impacts our

26:09

world the most? Is it in our personal

26:11

lives? Is it in our work lives? Is it

26:13

everything? We keep talking about what

26:15

it's going to mean for our healthcare

26:16

and medical world. Like where do you see

26:19

it impacting our world the most or will

26:21

it be everything? Well, I think there's

26:23

a number of places that it's already

26:25

impacting our world. But I think you

26:26

know if you if you take the the

26:28

discipline of software engineering for

26:29

example that's going through a lot of

26:31

change I think exciting change but also

26:33

there are people who are concerned

26:35

changes you know with that as well. I

26:36

think in the end it will be very

26:38

productive and we're going through that

26:40

same change at Google as well. But I

26:42

think if you look at every industry

26:44

there are amazing applications. If you

26:46

look at biotech there will be amazing

26:47

applications. If we look at uh as you

26:49

said healthcare you know there will be

26:51

amazing applications. I'm wearing my new

26:54

uh Fitbit Ace, which is a

26:56

>> I I spy three different tracking things.

26:57

>> I [laughter] am a wearable guy. I'm a

26:59

wearable guy. But I think the most

27:00

exciting part about wearables these

27:02

days,

27:02

>> that's the new screenless Google device

27:06

that from Fitbit.

27:07

>> Yeah. And I And I love it. And the the

27:08

the most exciting part for me is, you

27:10

know, we've had wearables, as you can

27:12

see, I've got a bunch of them, and I

27:13

I've been involved in the space for a

27:15

long time. They're collecting health

27:17

data, but it's what you do with the data

27:19

that's most important because, you know,

27:21

if you're looking at your sleep score, I

27:23

mean, I kind of know when I had a bad

27:25

night of sleep, you know, so the

27:26

question is what are the patterns? What

27:28

are the trends? What are the insights

27:30

that can really help with behavior

27:32

change? And I think that is where AI on

27:34

the device can really help us with that.

27:36

>> Yeah. If you're listening on radio,

27:37

Samir's got a Fitbit on one wrist. It

27:39

looks like an [laughter] Android watch.

27:40

>> I have a Pixel watching.

27:43

he's got to check out what the

27:44

>> competition.

27:46

[laughter]

27:46

So, I I I want to go back to this whole

27:48

idea of the ecosystem and and how it

27:50

relates to AI and Gemini and and and the

27:52

stickiness of it because I I I think

27:54

that's a a a part of the conversation

27:57

that really has has just recently

27:59

occurred to me and and I'm wondering

28:00

what advantage you think Google has

28:03

having Gemini as its own AI operating

28:07

system along with Android and that

28:10

working seamlessly across whether we're

28:12

talking about your Fitbit, whether

28:14

you're talking about an Android phone,

28:15

whether talking about a pixel watch uh

28:18

versus sort of third party LLM like uh

28:22

you know claude or chat GPT.

28:24

>> I want to say a couple things about

28:25

that. First um you know Android is an

28:27

open platform and so one of the things

28:29

that's the central tenant of Android is

28:31

choice. So Android is the only mobile

28:33

operating system where you can choose

28:35

your personal assistant. You can choose

28:36

the agent that's important to you. So if

28:39

you want Gemini, you can use Gemini. If

28:41

you want claude or something else, you

28:42

can choose that

28:43

>> and those will work equally well.

28:44

>> Yeah. The idea with Android is that it's

28:46

open and so you should be able to plug

28:48

in what's important to you. That's a

28:49

fundamentally important part of Android

28:52

and um and I think it's an advantage in

28:53

this world while things are moving fast.

28:56

From a Google perspective, of course,

28:57

Google competes with some of those

28:59

companies and I think Google has a

29:00

unique position in that it has a full

29:03

stack of of innovation that it's worked

29:05

on over over a decade from the from

29:07

silicon unique silicon in the data

29:09

center foundational models all the way

29:11

through to the operating system. And we

29:13

want to connect those things in ways

29:15

that are powerful and that help

29:16

consumers live uh live better lives.

29:19

That that's what we're doing. In fact,

29:20

recently we announced something called

29:21

Gemini Intelligence. Gemini Intelligence

29:24

is the best Gemini experience available

29:27

on the most premium Android devices. So

29:30

take a Samsung device or Pixel device

29:32

and that's really where you can

29:34

experience the best of that entire

29:35

stack. Do you think being open source

29:38

and being able to let users choose and

29:40

select what their agent will be will

29:42

give you guys longer term advantage?

29:45

Because I do think there's folks say

29:46

that at some point there's going to be

29:47

some shake out because we're not going

29:49

to need necessarily

29:51

every big LLM like at some point people

29:54

are going to financially we heard it

29:55

from the cerebras CEO last night that

29:58

people are going to select kind of what

30:00

they need and why.

30:01

>> I think that competition is good. Um and

30:03

I think that it's led to a ton of

30:05

innovation. I mean, look at all the

30:06

change we're seeing that's fueled by

30:08

everyone competing, which is great. From

30:10

an Android perspective, the thing about

30:12

Android being open source and being an

30:14

open ecosystem. It means that we work

30:16

with many, many different partners. We

30:19

work with, for example, of course,

30:21

Google builds its own hardware with

30:23

Pixel devices, but we also work with

30:25

Samsung and Motorola and many other

30:27

companies. And what that usually leads

30:28

to in times of very fast-paced change,

30:31

Android does better because that open

30:34

nature. you have innovation coming from

30:36

everyone and so you got foldable phones,

30:39

you've got different types of LLMs,

30:41

you've got different types of of of

30:43

screen technology and all of that is

30:44

great for the consumer and usually

30:46

pushes the ecosystem.

30:47

>> But it also

30:48

>> so yes, it's an advantage. [laughter] I

30:50

think so.

30:50

>> But it also sounds to me like it's a

30:52

it's a risk and it's a a branding risk

30:54

because the experience that someone

30:55

might get on a nonpremium smartphone or

30:57

a non-premium device that uses Android

30:59

might be completely different than the

31:00

experience that they're getting from one

31:02

of those high-end Pixel devices that you

31:04

talked about. So, how do you control for

31:05

that and how do you sort of protect the

31:07

brand in an open source world?

31:08

>> Well, it's a great question and in in

31:10

what what we do as Google, our role in

31:12

in this from an Android perspective is

31:14

to do a couple of things. first is to

31:16

make sure that the experience has a

31:19

certain amount of uh of compatibility.

31:22

Meaning if you get the Google Play Store

31:24

for example or any store of your choice

31:26

and you download apps, they should work.

31:28

They should work really well, right? So

31:30

that's a baseline set of compatibility

31:32

including the new LLM apps that that you

31:34

want to get. That's really important

31:36

that we create that baseline of

31:37

compatibility. But it also means

31:40

allowing manufacturers to bring the

31:42

latest and greatest technology. you

31:43

know, if we if we controlled everything

31:46

on our end, you wouldn't have foldable

31:48

phones. You wouldn't have flip flip

31:50

folds or or open folds and that's pushed

31:52

the industry forward. And so that is the

31:55

the dance. It's Android being open. You

31:58

want to create a great experience. You

31:59

also want to let innovation thrive

32:01

because that pushes the whole industry

32:03

forward.

32:03

>> Samir, you know, we've just got about a

32:05

minute or so left here and I look at

32:06

your background, Bachelor of Computer

32:08

Science from UC San Diego. You completed

32:10

a program for management development at

32:11

Harvard Business School. What's your

32:13

view on AI and education and maybe how

32:14

education has to evolve?

32:16

>> Yeah, that's a great question. Um, I

32:18

mean, I think that AI can play a huge

32:20

role for the positive in education if we

32:22

if we uh embrace it and use it

32:24

correctly. Obviously, there's a lot of

32:26

ways and a lot of concerns about misuse

32:28

of AI in in in education. I think that

32:32

the that you know my experience is that

32:35

there is not enough access to

32:37

highquality education around the world.

32:39

And uh if you if you think about for

32:41

example learning English um that's not

32:44

only a skill that is important if you

32:46

want to be bilingual in many countries

32:48

it improves your economic standing in

32:50

many places if you have if you're

32:52

multilingual or you or in particular if

32:54

you have learned English. Now how can we

32:55

get English language learning for

32:57

example to as many people in the world

32:59

as possible and there are companies like

33:01

open education and open English that are

33:03

doing exactly that using AI technology.

33:05

That's awesome. that provides wider

33:07

access and and I think we're going to

33:08

see a number of things like that which

33:10

help level the playing field in many

33:12

ways.

33:12

>> Um Samir, thank you so much. Give us so

33:14

much time. We really appreciate it.

33:15

[music] Samir Simat, president of the

33:17

Android ecosystem joining us here at

33:19

Bloomberg Technology.

33:21

>> Uh I love the the three different pieces

33:23

of wearable technology.

33:24

>> It's pretty impressive.

33:25

>> Yeah. [laughter] Are you getting

33:26

consistent scores, Samir?

33:28

>> No.

33:29

>> Okay. This is

33:30

>> No, I I have to say like the the Whoop

33:32

is the the excuse me, the the Ura is the

33:34

hardest on me. Um, and uh, but I do I

33:39

really love the new health coach with

33:42

Google Health. Um, I have like many of

33:45

us, I I'm sure we all have our I'm 48.

33:47

We all have our ailments. I have a lower

33:49

back. It's not great. I told it that. I

33:51

gave it my MRI report and it is totally

33:54

it has totally tailored my workout.

33:56

>> Yeah.

33:57

>> To prevent injury and like that's been

33:59

awesome.

34:01

>> Interesting stuff. Love it. Thank you so

34:02

much. I truly appreciate Simerat joining

34:04

us.

34:05

>> Stay with us. More from [music]

34:06

Bloomberg Business Week Daily coming up

34:09

after this.

34:13

>> You're listening to the Bloomberg

34:14

[music]

34:15

Business Week Daily podcast. Catch us

34:17

live weekday afternoons from 2:00 to

34:19

5:00 Eastern.

34:20

>> Listen on Apple CarPlay and Android Auto

34:22

with the Bloomberg Business App [music]

34:24

or watch us live on YouTube.

34:28

>> Well, let's continue talking AI and

34:30

specifically talk about Mythos. It's the

34:32

AI model from Anthropic that has central

34:34

banks, financial institutions,

34:36

governments around the world freaked

34:37

out. It's aimed at finding cyber

34:39

security vulnerabilities. Anthropic has

34:41

said it was too dangerous to make

34:42

available to the general public.

34:43

Actually, the co-founder and president

34:45

of Anthropic, Daniela Amade, was on

34:47

stage here earlier talking about the

34:48

model. If you missed any of that, you

34:50

can check out that interview at

34:50

bloomberg.com and on the terminal.

34:52

>> Yeah. And somewhere also seeing

34:54

opportunity. We should point out the

34:55

technology organization Mozilla, it is

34:57

known for Firefox, actually used um

35:00

Mythos to find and fix more than 271

35:03

vulnerabilities identified in a recent

35:05

version of Firefox. We want to get into

35:07

that and so much more. Joining us from

35:09

the Bloomberg Tech SE seg summit, excuse

35:11

me, in San Francisco's Rafie Coran. He

35:14

is the chief technology officer of

35:15

Mozilla. Welcome, welcome. Nice to have

35:17

you.

35:17

>> Thank you for having me.

35:18

>> You know, as we mentioned, people know

35:20

Mosilla as really the organization

35:22

behind Firefox. What are the

35:23

vulnerabilities though that Mythos

35:25

actually identified?

35:27

>> Yeah, I mean there's a bunch of them. A

35:28

lot of them had to do with being able to

35:30

break out of the browser itself and

35:31

actually execute code on your machine.

35:34

So like actually having JavaScript or

35:35

HTML come in and like just load it from

35:37

a web page and instruct the browser to

35:39

go do other things. So Mythos has been

35:41

very helpful for us to actually find

35:43

some of these really longstanding old

35:45

bugs and actually help us shut them

35:46

down. But what I noticed in the in the

35:48

blog post that that your team wrote

35:51

about this was that quote they wrote

35:54

quote we haven't we also haven't seen

35:55

any bugs that couldn't have been found

35:57

by an elite human researcher. So

36:01

was it as is it as powerful as everybody

36:04

says it is if if if a if a actual human

36:06

being could have identified these

36:08

things.

36:08

>> Yeah. No, I mean there's just a question

36:09

of just like how many elite human

36:11

researchers are there in the grand

36:12

scheme of things. So like, yeah, these

36:14

models are pretty powerful and they're

36:15

allowing us to do what like thousands of

36:17

these researchers could do instead of

36:19

just like six of them.

36:20

>> So would you have been able to identify

36:24

these with your own team?

36:26

>> We have a really good team. I think in

36:27

the fullness of time, yes, but like

36:29

allowing us to like rapidly accelerate

36:31

and just close out a bunch of stuff that

36:32

we didn't even know about.

36:33

>> So should we be freaked out about this

36:35

tech?

36:35

>> Yes. I actually think we should be. I

36:37

think we can like we'll be getting to a

36:38

world like once everything settles down

36:40

where things could actually be more

36:41

secure but like we're living in this

36:43

like intermediary zone. is the reason

36:44

why we should be scared because if

36:46

mythos is in the wrong hands they could

36:48

go to you know your bank and say uh okay

36:50

identify the vulnerabilities in you know

36:53

Rob's account and then

36:56

>> they get access to your money

36:57

>> the bank the water company the power

36:59

company like all the critical

37:00

infrastructure of the world like I

37:02

actually have faith that like internet

37:03

folks will figure this out like we're

37:05

used to patching things and deploying at

37:06

scale like heart bleed when SSL had

37:08

problems

37:09

>> but I don't think my bank knows how to

37:11

do it like you said I don't think my

37:12

power company knows how to do But those

37:14

are like the other pieces of critical

37:15

infrastructure that I'm really worried

37:17

about.

37:17

>> I mean, you're CTO. You should tell

37:18

those CTO. Okay.

37:20

>> So, there are things that are happening.

37:23

Why isn't it happening more or is it and

37:25

where they're being caught or what?

37:26

Like, give us an idea.

37:27

>> Yeah. No, I think we need to have

37:29

actually a real large scale like Y2K

37:31

like effort to actually start closing a

37:33

bunch of these critical vulnerabilities.

37:35

And you know, many kudos to what

37:36

Anthropic is doing, but we need to give

37:38

access to more people like we need to

37:40

get access to all the open source

37:41

providers. We need to get access to all

37:43

the database providers and then we need

37:45

to fund them to actually start doing the

37:47

work. So not only do you need to clean

37:48

the bugs up, but then we also need to

37:50

start thinking about mitigation. Like

37:51

this is actually a large scale. Look,

37:53

think about how much work we did for

37:54

Y2K. Boards got involved, governments

37:57

got involved, insurance got involved. We

37:59

haven't kicked all those in yet.

38:00

>> Why aren't we doing that yet? Because

38:02

this is moving so fast. Is it just Is

38:04

that it?

38:04

>> That's exactly the problem. I think this

38:06

is just moving so fast. like Mythos was

38:08

only really given to us in the like in

38:10

the in the winter time. It's only been a

38:13

couple of months, but the real question

38:14

is that open models are catching up. So

38:16

like I think it's only 6 to9 to 12

38:19

months when all the open models can

38:20

start doing what mythos are doing. So we

38:22

only actually have a small window of

38:23

time. So then how do we get to a point

38:27

where you know you're sitting here with

38:28

us a year from now?

38:31

>> Mythos is going to be old news. Y

38:32

>> there's going to be some sort of next

38:34

model that is just that much better. How

38:37

do you how are you going to be sure that

38:39

it's a company that's you know quote

38:40

unquote responsible behind it like many

38:43

would say anthropic is being with this

38:45

roll out maybe too responsible in your

38:47

view um but how do you make sure that

38:49

it's not a bad actor who has this done

38:52

yeah no I mean I think this is actually

38:54

where some government intervention might

38:56

not be a bad thing like actually making

38:58

sure that we put the right investment

38:59

into actually getting these things shown

39:01

deployed exactly the right people

39:02

>> you been to the post office

39:03

>> yeah I know 100% like

39:05

>> I mean We but we do need to do all this

39:08

work across all the efforts. We need to

39:10

secure that post office too.

39:11

>> I don't disagree with you, but I think

39:13

there are many people out there who say

39:14

this could be something that maybe the

39:16

free market would solve more uh

39:19

efficiently than government bureaucracy.

39:21

You have experience in in in in

39:23

government sort of. So

39:24

>> no 100%. I mean

39:25

>> politics you have experience.

39:26

>> I think the real problem is that public

39:27

sector infrastructure is not treated in

39:29

the same way. And so like we really need

39:31

to figure out how to do the real

39:32

investment there. And I don't think it's

39:34

purely just maybe a public private

39:36

partnership solves that, but I don't

39:37

think that's purely a market problem.

39:39

>> Rafie, you were the first CTO of the

39:40

DNC, the Democratic National um

39:42

committee, politics today, we see the

39:46

impact social media,

39:48

>> some goods, some bad.

39:49

>> Sure.

39:50

>> Um on politics, what about AI? No, I

39:53

mean like I think about this a lot like

39:54

we living in this world where we're like

39:56

owners not renters on all this

39:58

technology like this techn like just

40:00

think about your search experience. It

40:01

used to be 10 blue links on Google and

40:03

now it's one result when I ask a chatbot

40:06

and it's a right result for somebody

40:08

like is it right result for me? So I

40:10

really worry about this world where like

40:12

living in someone else's intelligence

40:14

all the time what does that actually do

40:15

and how we think how we decide how we

40:17

buy all those kind of things. Can we

40:19

actually get ahead of that and be better

40:22

with it?

40:22

>> I mean, this is where I really think we

40:24

need to figure out how to do open

40:25

deployments and open governance. Like

40:27

right now, the incentive models are just

40:29

such that like there was a whole study

40:31

based out of I think University of

40:32

Maryland where they just asked a bunch

40:34

of these chatbots to help you buy

40:36

something and they're vastly weighted

40:38

toward sponsored and promoted goods. So

40:41

like how do we make sure that they're

40:42

really on my side? I need to be able to

40:43

go into this, understand it, tweak it,

40:46

work with it as opposed to just being

40:47

delivered it. So we need to start

40:49

working in those kind of lexes. Now

40:50

>> you you were talking a little bit about

40:52

your family earlier and I'm just

40:54

wondering how how you as a CTO

40:57

>> and somebody who understands this

40:58

technology and the potential downfall

41:00

and and pitfalls of it rather you sort

41:02

of have that diet that information

41:05

technology diet.

41:06

>> I mean can I tell you the pros and cons?

41:08

>> Like my 10-year-old is vibe coding video

41:11

games like they're actually unbelievable

41:13

cute and stinky production. They're like

41:15

it's [laughter] like the best thing

41:16

ever. Um, and this other is an

41:18

expression is creativity. My

41:19

13-year-old, I got a phone call from a

41:21

teacher saying, "I think you use catch

41:22

GPD to cheat on an essay." So, like, we

41:25

do need to figure out like how we

41:26

actually treat people. How do we do

41:28

teach critical thinking? Like, that's

41:30

the thing that's still missing. And

41:31

that's the thing I think we need to

41:32

really home in on. It's less about the

41:34

tool, it's about that.

41:35

>> So, how do you do it though?

41:36

>> I think it's asking real questions. I

41:38

think it's having lots of conversations.

41:39

You know, the pope is in cyclical called

41:42

it fasting from AI. So, like we just

41:44

need to figure out a way like we need to

41:45

teach them this new tech, but we also

41:47

need to just monitor and help them like

41:49

do productive struggle.

41:50

>> Um, we're being told we have one more

41:52

question with you and it's Twitter or

41:54

Uber.

41:55

>> So, go I'm going to say Tim's got to

41:57

ask.

41:57

>> Yeah. I mean, you were you were at

41:59

Uber's advanced technology department.

42:01

This was like the precursor to

42:03

self-driving cars. It was This

42:06

department was killed a few years ago.

42:08

The promise was that you would have

42:10

driverless cars at at Uber.

42:13

the technology as it is today. I mean,

42:14

you walk outside of here, there are

42:15

Whimos going everywhere, not really in

42:18

New York City at this point. Where are

42:20

we when it comes to just getting into a

42:22

car and it taking us where we need to go

42:23

and there's no steering?

42:24

>> I mean, it turns out to be a really hard

42:26

problem. This is one of the things we

42:27

learned, right? It turns out to be a

42:28

really hard problem to do generalized

42:30

self-driving like to be able to drive

42:31

anywhere under any condition, under any

42:33

situation like a city-byc city basis. It

42:36

seems to be working out like Whimos are

42:38

pretty good. You know, I recently

42:40

crashed my Tesla under full

42:42

self-driving. You did? I did. So, like

42:44

the Teslas clearly had problems. We

42:45

slammed into the wall at like 30 miles

42:47

an hour. Oh my god. And so, like we

42:49

there are clearly still problems with

42:50

all this technology, but

42:52

>> that doesn't sound like full

42:53

self-driving.

42:53

>> It doesn't sound like full self-driving.

42:55

I totally think that the Tesla

42:56

technology is not quite ready yet.

42:57

Whereas Whimo has been designed to

42:59

literally not have a purse in the wheel.

43:01

The whole human in the loop version of

43:02

self-driving I think is not the right

43:04

path that we should be going on and we

43:05

should be thinking about just like

43:07

actually tackling the real problem head

43:08

on.

43:09

>> 20 seconds left. What's the question we

43:11

should all be asking ourselves in this

43:12

environment today?

43:13

>> Yeah, I mean the question is like how do

43:14

we make sure all these technologies are

43:16

deploying are actually on our side not

43:18

on someone else's side like I want to

43:19

live in a world of 7 billion AGI not

43:22

seven AGIS. How do we get there?

43:23

>> Ah really interesting Robbie thank you

43:25

so much.

43:25

>> This was a lot of fun. Thank you.

43:26

>> Yeah fun for us too. Robbie Coror and

43:28

he's chief technology officer at

43:30

Mozilla.

43:31

>> Stay with us. More from Bloomberg

43:33

Business Week Daily coming up after

43:35

this.

43:39

>> You're listening to the Bloomberg

43:40

Business Week Daily [music] podcast.

43:42

Catch us live weekday afternoons from

43:44

2:00 to 5:00 Eastern.

43:46

>> Listen on Apple CarPlay and Android Auto

43:47

[music] with the Bloomberg Business App

43:50

or watch us live on YouTube.

43:53

Carol Master Stick Life here at

43:55

Bloomberg uh technology out here in San

43:58

Francisco getting ready to wrap up, but

43:59

we have one more important conversation

44:01

to bring to you.

44:02

>> Yeah, we're very excited to have Katie

44:03

Han with us. Uh you know her because

44:06

she's a former federal prosecutor. She

44:07

was the first female general partner at

44:09

Andre Horowitz. Uh she made her name

44:12

investigating digital assets before

44:13

investing in crypto startups. She's

44:15

raised a billion dollars for new new

44:17

venture funds. She's also getting ready

44:19

to to join Bloomberg's Emily Chang up on

44:21

the main stage to talk about where she's

44:23

looking when it comes to new

44:24

opportunities in tech and she's joining

44:26

us on Bloomberg Business Week Daily.

44:28

>> We're like eager to pick her brain.

44:30

Katie Han is founder and CEO at Von

44:32

Ventures. Welcome. Welcome. Thank you

44:33

for joining us.

44:34

>> Thank you. Welcome to California.

44:35

>> Thank you. We love California. Um what

44:38

do you when you look at the world

44:40

broadly? We're having so many

44:41

conversations about AI. We're going to

44:42

talk crypto and some other things, but

44:44

where do you think about the best

44:46

opportunities and where you might want

44:48

to commit investing?

44:49

>> Well, I think how we think about it is

44:51

we've raised a billion dollars in new

44:53

funds. Um, and the world looks really

44:54

different than four years ago when we

44:57

raised our first fund. And it even looks

44:59

different than 12 months ago. And what

45:01

we're doing and where we're excited to

45:02

invest, um, where we're seeing a lot of

45:04

opportunities is what I'm calling the

45:06

new economy. It's founders that are

45:09

building the future of finance, for

45:10

example. And obviously digital assets

45:12

are a part of that. But we're seeing

45:14

kind of three structural shifts right

45:15

now that we're keeping a close eye on.

45:17

The first structural shift we're seeing

45:19

is what I'll call the new plumbing or

45:20

the new financial infrastructure. This

45:23

is financial infrastructure that's built

45:25

for a global world from day one. It's

45:27

built for a 24/7 and it's built for a

45:29

digital world. So that's the first

45:30

structural shift. The second structural

45:32

shift is kind of new asset classes

45:34

entirely and new markets and it started

45:37

as stable coins a new digital form of a

45:39

dollar for example but quickly we're

45:41

seeing tokenization of stocks of bonds

45:44

and of other funds um and so with those

45:46

new kind of asset classes also you can

45:48

have new markets things like prediction

45:49

markets which we can talk about things

45:51

like perpetual market um that's getting

45:54

a lot of attention ahead of the SpaceX

45:55

IPO and then the third structural shift

45:58

we're seeing and this is really exciting

45:59

and probably the furthest way is what

46:02

I'm calling the future of agentic

46:03

finance and it is quite simply what does

46:05

the world look like when you're building

46:07

financial services and products for the

46:09

end user who's not a human but the end

46:12

user who is an agent or a computer and

46:14

that's just kind of a mind-blowing world

46:16

what does that look like they don't stop

46:17

working by the way uh they're 24/7 so

46:20

it's got to be always on

46:22

>> is is crypto still though part of the

46:24

DNA of the fund

46:25

>> absolutely you know we've always said

46:27

we're we're looking to back founders who

46:29

are building the future of finance and

46:31

cryptographic cryptographic primitives

46:34

and digital assets are still a huge

46:35

piece of that. I mean, stable coins, a

46:37

lot of people don't even think of stable

46:39

coins as as crypto. Um, but I think they

46:42

very much are. So, is the tokenization

46:44

of assets, like I said, prediction

46:46

markets, but but there's a lot of

46:48

synergies between

46:49

>> you've had a really strong track record

46:50

with stable coin companies. I mean, we

46:52

can just go through some of these.

46:53

Bridge uh Stripe acquired for $1.1

46:55

billion. That was a home run for you

46:57

guys. Um, you also invested in the

47:00

stable coin company BVNK, sold a

47:02

Mastercard for about $1.8 billion. So,

47:04

yeah, you know what you're talking about

47:06

when it comes to stable coins, but is

47:08

that era like are the opportunities

47:11

for early stage venture capitalists or

47:13

are we past that because of that? It all

47:16

depends on how you define early stage,

47:18

but I think we're still in the early

47:19

innings. You know, actually, I've been

47:20

talking about stable coins since 2017. I

47:23

mean, I debated Paul Krugman about the

47:25

promise of stable coins in 2018. Um, and

47:28

he said it was a flash in the pan and

47:29

obviously I think u, you know, we're

47:31

seeing that it's not. I mean,

47:32

Mastercard, the the transaction you just

47:34

mentioned, it was the third largest

47:36

acquisition in history that company's

47:37

ever done.

47:38

>> Stable coin is not a a flash in the pan.

47:40

I think that's fair to say.

47:42

>> Bitcoin right now is under serious

47:44

pressure.

47:45

>> Having a moment. Yeah, having a moment.

47:47

>> We're we're, you know, we're basically

47:48

at levels not seen since 2024. really

47:52

the

47:52

>> well we saw them in February.

47:54

>> We did that's what I mean like so

47:55

February so like we're down to those

47:56

levels.

47:57

>> Um but my point is that they haven't it

48:00

hasn't participated in the risk on rally

48:02

that we've seen. I mean the NASDAQ 100

48:03

is up more than 30%.

48:05

>> Just since the the end of March

48:07

>> Michael Sailor's strategy is selling

48:09

just every so little bit of Bitcoin.

48:12

What's what's going on right now? I I

48:13

logged in X earlier this week and it

48:14

seems like people are losing faith.

48:16

>> Yeah, I don't think so. I think, look, I

48:17

I remember where I was the day Bitcoin

48:20

first hit $1,000. I remember where I was

48:22

the first day it hit $10,000. And I

48:24

remember where I was the day it first

48:26

hit $100,000. And I wonder where will I

48:28

be when it first hits $200,000. I think

48:30

it's just a question of time horizon.

48:32

And of course, we're not a hedge fund.

48:33

We're not trading in and out of of these

48:35

assets. We are a long-term venture

48:37

investor. Um and and you know, we've

48:39

made bets in our in our first fund in

48:41

Bitcoin, for example, and those have

48:42

done very well for us. So, I'm still a

48:44

believer in the long-term store of value

48:46

story that Bitcoin, it clearly has

48:48

product market fit. Another thing I

48:50

debated Paul Freriedman on years ago

48:52

when it was uh in the billions of

48:54

dollars as an asset class and now today

48:56

uh that number is in the trillions. And

48:58

again, to your point, it does vary

49:00

dramatically. But if you would have told

49:02

me a few years back has crashed or have

49:05

a moment at 65,000 or 64 63,000, I would

49:09

have told you, oh my gosh, really

49:11

incredible. Um, so it's all about time

49:13

horizons and that's why we're taking a

49:15

long-term patient view.

49:16

>> Katie, is AI sucking some of the

49:18

interest out of the retail investors or

49:20

retail trade and they're kind of

49:21

shifting their their attention?

49:23

>> Yeah. Well, you know, one of the

49:25

interesting things about the digital

49:26

asset [clears throat] space that I've

49:27

been really really pleased to see is the

49:30

institutional story. And I think that's

49:32

the quiet story and that's what's really

49:33

changed. So where you have retail

49:35

focused is not necessarily always where

49:37

you have institutions focused. And it

49:39

used to be the case that retail made up

49:42

90% of kind of the digital asset

49:44

economy, 10% institutions. Now that

49:46

number has shifted. It's it's kind of

49:48

closer to 20 even 30% institutional. And

49:51

I think I would actually say that's why

49:52

we're seeing the volatility lessons. I

49:54

know it doesn't feel like it from what

49:55

you're seeing.

49:56

>> It was more tradition like that's right.

49:58

It used to be much more retailheavy and

50:00

and still retail is a very important

50:02

component of the digital asset story,

50:04

but it's not the only component. And I

50:05

think what we're seeing is we're seeing

50:06

a lot of institutional participation.

50:08

You're seeing Black Rockck. I mean, you

50:10

mentioned Michael Sailor, but the the

50:11

reality is that um you know, it was a

50:13

very dimminimous amount that he sold and

50:15

it's all about the net accumulation.

50:17

>> Yeah. I think for some people it was

50:18

symbolic because he's such a maximalist.

50:20

>> Absolutely. Absolutely.

50:22

>> I want to talk some personal investments

50:23

versus investments in the fund. I mean,

50:26

we go through the list of like some of

50:27

the hottest companies. You've invested

50:28

them in them personally. um Cerebrris,

50:31

Anthropic, Anderil, Coinbase, Data

50:33

Bricks, Cognition, Figma, Gro, Nico,

50:36

Open AAI, SpaceX, the list goes on. How

50:40

do you separate sort of your interests

50:42

as a as a as an investor personally?

50:45

Yep.

50:45

>> Versus the fund because I'm sure there

50:47

are a lot of people who invested in your

50:49

fund who'd say, "Wait, I would love to

50:50

have investments on that list."

50:52

>> And I think many of them do have

50:53

investments on that list. Um the me the

50:56

thing that you mentioned, it's very easy

50:58

actually. If you're in the future of

50:59

finance, if you're in digital assets, um

51:01

the new economy, very clearly it goes in

51:04

the fund and the fund has the first

51:05

opportunity. But if if you're trying out

51:07

you you mentioned Neco, you mentioned

51:08

Edison, these are great examples. These

51:10

are bio companies, these are longevity

51:12

companies and they're really harnessing

51:14

the power of AI. I mean, if you take a

51:16

company um like Edison, it's doing drug

51:18

discovery and shrinking the discovery of

51:20

medicines using AI. That's clearly not

51:23

in our mandate as a fund of the future

51:25

of finance. But I'm really lucky being

51:27

where I am and being in these um circles

51:29

to meet these incredible founders. Neco,

51:32

another one, you know, it's coming to

51:33

the US. People who are in London have

51:34

obviously heard of Neko. That's Daniel

51:36

Ek uh one of his one of his new new

51:39

ventures,

51:40

>> founder of Spotify.

51:41

>> Founder of Spotify. It's really exciting

51:43

what they're building. Um what he and

51:44

Shaq are building with Neco. Um, and

51:46

there's a long waiting list in London,

51:48

but it's coming to New York, actually,

51:49

to your neck of the woods later this

51:51

year, and it's going to be under $500.

51:53

And you're really talking about

51:54

preventative medicine that is

51:56

accessible. And that $500 is the

51:57

starting point. Obviously, I think that

51:59

company, you can talk to them, helps to

52:01

bring that price down over time. You're

52:03

talking about really harnessing the

52:04

power of technology for preventative

52:06

medicine for everybody. Uh, not just for

52:08

a few people. And I think that's really

52:09

powerful. And it's it's not in our fund

52:11

mandate, but I'm not going to stick my

52:12

head in the sand and not participate in

52:14

those. And it's really great for our

52:16

founders to be in touch with founders of

52:18

the Han Ventures portfolio because

52:20

there's actually a lot of synergies

52:21

between some of these technologies.

52:22

Well, I love that you went there cuz

52:24

when it comes to AI, I mean, there's a

52:26

lot of obvious things that we think it's

52:28

going to do, but I think about the

52:30

things that you guys might be investing

52:32

in that maybe isn't in the headlines or

52:34

the main stories. I mean, where do you

52:35

think investors need to be thinking

52:37

about how this ultimately impacts our

52:39

world?

52:39

>> Well, look, there's two areas that our

52:41

fund is is really excited about. you

52:43

asked if it's sucking all the air out of

52:44

digital assets and I would say no

52:46

there's so much synergy between the two

52:47

frontier technologies uh if you think

52:50

about like cryptographic primitives and

52:51

you think about AI one of the things I

52:54

think about is a commerce right and we

52:56

have companies like Stripe Visa

52:58

Mastercard Coinbase looking at agentic

53:01

commerce platforms and if you think

53:02

about I think we can all agree can't we

53:05

that agents are going to be doing more

53:06

than humans are going to be doing with

53:08

the with with these tools right um who

53:11

knows about the pace of that who knows

53:13

how much work they'll take from humans,

53:14

but certainly I think we can all agree

53:16

that computers are going to increasingly

53:18

take on more and more of our work and

53:20

and they're going to need to pay for

53:22

things if they're going to be effective.

53:23

They're going to need to transact and

53:25

they're going to need to operate 24/7

53:27

and we don't think they're going to use

53:28

the old payment rail to do that, right?

53:30

They're going to have to be able to

53:31

subscribe to services, right? And and

53:33

what are they going to use? We think

53:34

they will use technologies like stable

53:36

coins. Another really important thing

53:38

I'm really excited about, four words,

53:40

provenence, identity, privacy,

53:42

reputation. Those are so important in

53:44

the age of AI. And these are things that

53:47

people in cryptography have been working

53:49

on for years.

53:49

>> Do you have agents acting on your behalf

53:51

right now?

53:51

>> Yeah, we do at the funds. We do at the

53:53

fund. Um well, I'm not going to give

53:54

away all of our secret sauce here uh to

53:56

folks who are are are looking, but we're

53:58

we're experimenting. I think like many

54:01

um you know I I still think we're at a

54:03

point where we still need a lot of human

54:04

oversight and particularly I do have

54:06

people who tell me well AI agents can

54:08

pick and they're better pickers than

54:09

humans. I don't believe that for a

54:11

second right now. I still think there's

54:12

a really important role for humans to

54:15

play especially when you're judging

54:16

founders and figuring out what to invest

54:18

in. Right.

54:18

>> Well Samir Samad over at Android

54:20

Ecosystem gave us a great example of how

54:21

he was, you know, having an agent help

54:23

him get food for an Instacart for a

54:26

barbecue that he was having.

54:27

>> Sure. Do you want an example? I mean,

54:29

we're definitely using it to help us

54:31

make sure our pipeline is strong. Are we

54:33

seeing everything? Um, definitely my

54:35

investment team is using cloud and using

54:38

codeex and using other technologies to

54:40

make sure we're scanning the full

54:41

surface because of course where we're

54:43

investing is not just here in Silicon

54:44

Valley, it's global. Katie, what you're

54:46

seeing and maybe and I think we've

54:48

probably got to wrap up soon, but in

54:49

terms of dislocation to the economy and

54:51

jobs. Yeah.

54:52

>> Is it just another technology where

54:54

there's going to have to be a little

54:56

adjustment period or is it something

54:57

more significant based on what you're

54:59

seeing?

54:59

>> Look, it is transformational technology.

55:01

I think what we could debate is the

55:03

timeline. Okay.

55:04

>> Uh and I think that's really unknown. I

55:06

think there are a lot of people that are

55:07

prognosticating that that's going to

55:08

happen this year. I think that's really

55:10

unknown um as to how long that's going

55:13

to take. Um, and I still think there's a

55:15

really important role for humans to play

55:17

here. But I do think it it is something

55:19

serious to pay attention to just

55:21

socially and economically. Um, and one

55:23

of the things I think back to is, you

55:25

know, the computer era. Obviously that

55:27

replaced a lot of jobs, but it also

55:29

created a lot of jobs. So I think

55:30

there's truth on both sides and I think

55:31

it's something everyone all of us need

55:33

to pay close attention to.

55:35

>> Great stuff. Um, stay in touch. Thank

55:37

you so much as you I love also to going

55:39

into the medicine and and longevity

55:41

because I think these are things areas

55:42

that could be impacted um greatly.

55:44

Katie, thank you.

55:45

>> Thank you for having me.

55:45

>> Have fun on the stage.

55:47

>> Good to see you.

55:48

>> Take care. You too.

55:49

>> Katie Han, founder and CEO over at Han

55:50

Adventures.

55:51

>> This is the Bloomberg Business Week

55:53

Daily podcast available on Apple,

55:56

Spotify, and anywhere else you get your

55:58

podcasts. Listen live weekday afternoons

56:01

from 2 to 5:00 p p.m. Eastern on

56:04

Bloomberg.com, the iheartradio app,

56:07

TuneIn, and the Bloomberg business app.

56:09

You can also watch us live every weekday

56:11

on YouTube and always on the Bloomberg

56:14

terminal.

56:17

[music]

Interactive Summary

The video features a Bloomberg Tech summit in San Francisco, covering key discussions on artificial intelligence, technological innovation, and its impact on the economy and society. Brad Stone discusses the current AI hype cycle, comparing it to the late 90s internet boom, while highlighting the tension between optimism and safety concerns. Samir Samat of Android explains the transition from an operating system to an 'intelligent system' powered by AI, and Rafie Coran of Mozilla talks about using AI to identify security vulnerabilities. Finally, Katie Haun discusses investment opportunities in the 'new economy' and the synergy between AI and digital assets.

Suggested questions

4 ready-made prompts