0:04

Welcome to Big Questions. This

0:07

is Cal Busman, lots to

0:09

talk about this week, starting

0:11

with how much information you

0:13

can give away about yourself

0:15

in an hour. Simply by

0:17

getting up in the morning,

0:19

going on the computer, doing

0:21

a few Google searches, sending

0:23

out a few text messages,

0:25

heading out to a Starbucks,

0:27

using a credit card to

0:29

pay for your coffee, and

0:31

walking down the street where

0:33

cameras are videoing you and

0:35

capturing your face so it can

0:37

recognize it. Episode moves on

0:39

to the algorithms that are sending

0:41

you what they think you

0:43

want to see on your phone

0:45

and computer so that you'll

0:47

keep coming back for more. And

0:49

how that separates us from

0:51

a time where many of us

0:53

could see the same things

0:55

as those around us and have

0:57

a collective reference point. Will

0:59

we all eventually see different realities

1:01

and be turned into a

1:03

universe of one? And this moves

1:05

on to the place in

1:07

the future where we'll have

1:09

chips implanted in our brains

1:11

and won't need to talk.

1:13

We'll think a question and

1:16

the answer will come back

1:18

to us. So welcome to

1:20

this week's episode of Big

1:22

Questions with award -winning professor at

1:24

Columbia University Business School, Sandra

1:26

Matz, who's written a book

1:28

called Mind Masters, the data

1:30

-driven approach of predicting and

1:32

changing human behavior. In fact,

1:34

it just came out today. If

1:36

you want to get an idea

1:38

of where the world is

1:40

headed, just listen to Sandra talk

1:42

about how our digital footprints

1:44

are being used to understand us

1:46

even better than we understand

1:48

ourselves. She'll show you where all

1:50

this has taken us and

1:52

offer suggestions on how we might

1:54

make our data work best

1:56

for us and society. So here

1:58

we go from A small

2:01

town in Germany

2:03

to Columbia University

2:05

in New York,

2:08

let's get straight

2:10

to Sandra Motz

2:13

and your future.

2:15

Sandra, it is a

2:17

delight to be with you

2:19

after going through your book

2:21

and... Then just talking to

2:23

you for a few minutes,

2:26

you came up with an

2:28

astounding fact that just shows

2:30

how many data points we

2:32

create that other people have

2:34

access to. We are probably

2:36

completely unaware of we may

2:38

be aware of some, but

2:40

you know, your book begins

2:42

with you taking a walk

2:44

in chapter one on early

2:46

morning, and like an hour

2:48

later. you return and you

2:50

point out that six gigabytes

2:52

of data has been amassed

2:54

about you and what I

2:56

want to know is all

2:58

right how could we see

3:00

that like you use these

3:02

phrases gigabytes terabytes very hard

3:04

to like understand what this

3:06

means how many of these

3:08

little data points that we're

3:10

giving away without even knowing

3:12

we're giving them away. Is

3:14

there a way for you

3:16

to clarify it so somebody

3:18

can kind of get a

3:20

grip on it? Rub their

3:23

head around and happy to

3:25

and I thought it was

3:27

mind-blowing by the way. So

3:29

first of all it's it's

3:31

a lot of zeros, right?

3:33

Six zeros. And we just

3:35

did the math and essentially

3:37

every hour you and I

3:39

generate more data than the

3:41

spaceship that we used to

3:43

launch and people into space

3:45

and land on the moon

3:47

had in terms of memory.

3:49

So it's like 500 million

3:51

times more data that we

3:53

generate every 60 minutes than

3:55

was on this chip that

3:57

flew the spaceship into into

3:59

space. So to me, that's

4:01

pretty mind-blowing. What

4:03

does that mean for us?

4:06

Because we really don't know where

4:08

this is, that is going. At

4:10

least we knew back in 1969

4:12

it was taking us to the

4:14

moon. To the moon. But we

4:17

have no control over understanding

4:19

this. Where is it all

4:21

going? Like you start the

4:23

book with basically an hour

4:25

in your life that just seems...

4:28

you know, very ordinary. And then

4:30

you come up with this figure

4:32

that is to me, not ordinary at

4:35

all. For me, it's not even just

4:37

a figure, right? So a figure is

4:39

a figure and yeah, we do

4:42

generate a lot of data. For

4:44

me, it's just the fact that

4:46

it's incredibly rich data, right? So

4:48

forget about the volume, the fact

4:50

that... based on all the Google searches

4:52

that you make, right? You wake up, you

4:55

check your phone, so first of all I

4:57

know that you're up now because your phone

4:59

is unlocking, and I can see who you're

5:01

texting, I can see which websites you're browsing,

5:03

which news you're reading, then if you wake up

5:05

and go get some breakfast somewhere, I

5:07

can see that you swipe your credit

5:09

card, which means that I know what

5:11

you're purchasing, I know exactly where you

5:13

are, potentially know who you're meeting with,

5:15

because that's two phones showing up in

5:18

the same location. And even if you

5:20

don't have your credit card or your

5:22

phone, there's cameras in the streets pretty

5:24

much everywhere, capturing with facial recognition that

5:26

you've been there at a certain point

5:29

in time, we obviously have wearable devices

5:31

these days. So there's pretty much no

5:33

way that you can escape this data

5:35

tracking. And for me, what I think

5:38

is so interesting about this data is

5:40

that when you put these puzzle pieces

5:42

together, it's not just individual data

5:44

points, right? If I think about,

5:46

well, someone knows that I bought... coffee

5:48

lathed Starbucks. So be it, right? It doesn't

5:50

seem like the end of the world, but

5:53

what we've seen over the last really like

5:55

10, 15 years now in terms of science

5:57

is that we can put these puzzle pieces

5:59

together. at a really interesting and

6:01

accurate reading of who you

6:03

are on the inside. So

6:06

your psychology, anything from personality,

6:08

mental health, like your values,

6:10

your moral values. And that

6:12

to me is both fascinating

6:14

and terrifying because as you say

6:16

it's out there. So that's one very

6:18

big question I got about all

6:21

these algorithms. And I'll just introduce

6:23

the second question now and maybe

6:26

we could just frame. the podcast around

6:28

these two questions and also

6:30

I know from your book

6:32

that you do offer up

6:34

some solutions or attempts at

6:36

solutions. So maybe that one two

6:39

three is a nice way to

6:41

frame the whole podcast. So number

6:43

two to me what I'm noticing

6:45

is that I'm in conversation

6:47

with AI every day. I'm asking it

6:50

to do a lot of

6:52

things. It's very friendly to

6:54

me in fact. That's somebody

6:56

on the podcast who told

6:58

me, treat it warmly like

7:00

you would a person. And

7:02

it will actually do better

7:04

work for you. I don't know if

7:07

you believe that. But it's

7:09

interesting to test that.

7:11

Yeah. It's interesting because there's

7:14

a lot of pleasantry. It

7:16

seems like the AI cares

7:18

about me. Yeah. Just because

7:20

it's learned language model is

7:23

told it. Okay, when somebody

7:25

treats you like this, it's

7:27

good to trade them back like

7:29

that. It's certainly mimicking. So

7:32

that's probably part of

7:34

it, right? Same way that

7:36

we humans do. Right. So I'm

7:38

getting closer and closer to

7:40

AI. And what I'm noticing is

7:42

generally when I go on the internet,

7:45

just say I go on X or

7:47

Twitter as it was once known,

7:49

it is sending me... material,

7:51

stories, content that it knows that

7:53

I'm going to be interested

7:55

in. And I'm just assuming

7:57

that everybody's getting...

8:00

the same thing. So everybody

8:02

is getting only what they

8:04

particularly are interested in. And

8:07

so we're all seeing different

8:09

things, which is very different from

8:11

my childhood when we all turned

8:13

on the television at a certain

8:16

time in the evening and we

8:18

all got the same news. So

8:20

we all could look at something

8:22

and even if we had very

8:25

different opinions of it.

8:27

We couldn't have a different

8:29

opinion about what was told to

8:31

us. But now everybody's

8:33

getting a different opinion.

8:36

And so there's just

8:38

less and less connectivity

8:40

that we can have because I'm

8:42

saying something that you're saying.

8:44

There's no sameness to latch

8:46

on to. And it's not just news,

8:49

right? It's like what I think

8:51

is interesting too, is it's also

8:53

all of the cultural. icons.

8:55

Like we used to have

8:57

fun conversations because we watched

8:59

the same shows and we saw

9:02

the same stuff. It's so I think

9:04

this is like part of it,

9:06

but continue. No, go ahead.

9:08

I'm very interested about this

9:10

because it's those cultural

9:12

icons that keep us together.

9:14

And I know I'm going backaways,

9:17

but actually in the 60s, in

9:19

early 70s, there was a show

9:21

in America called Laughin. It was

9:23

a comedy show. And because

9:25

there are only three major

9:27

networks, it basically got 50%

9:29

of the market. Because it

9:31

was good, everybody watched it.

9:34

And when there would be a

9:36

comedy bit, and the character would

9:38

say something like, here comes the

9:40

judge. Like the next day,

9:43

everybody knew what was. Yeah, exactly.

9:45

Here come the judge. Here come

9:47

the judge. And we have had

9:49

these. At the time, I don't

9:51

want to call him iconic because

9:53

it didn't really last long. Nobody

9:56

would remember here comes judge unless

9:58

you're like over 60 now. But

10:01

the fact is that

10:03

there were things beyond the

10:05

news that kept us

10:07

together. And I just wonder

10:09

if we're going to

10:11

have that or if this

10:13

just keeps ramping up.

10:15

When I say ramping up,

10:17

I've been thinking of

10:19

some big numbers because I

10:21

saw that Google revealed

10:23

a computer chip that could

10:25

basically in five minutes

10:27

solve a problem that the

10:29

supercomputer now would need

10:31

10 septillion years. I don't

10:33

think you can even

10:35

wrap your head around those

10:37

numbers. That's many, many,

10:39

many more zeros than we

10:41

talked about in the

10:43

beginning. Yeah, I actually asked

10:45

AI about it. And

10:47

it told me that if

10:49

I'm counting, if I

10:51

want to count to a

10:53

million, one, two, three,

10:55

it will take me 12

10:57

days. If I want

10:59

to count to 10 septillion,

11:01

it would be like

11:03

counting from the time that

11:05

the universe was born

11:08

till now, times 23. Times

11:10

one, and a little

11:12

bit more. So you can

11:14

see how fast things

11:16

are going to keep moving,

11:18

keep moving, keep moving.

11:20

And I wonder if that

11:22

speed is also going

11:24

to keep splitting our connectivity

11:26

so that we're going

11:28

to be relying less and

11:30

less on other people.

11:32

And if AI starts taking

11:34

jobs that we were

11:36

already shifting to at homework,

11:38

you know, the fewer

11:40

places that you have, whether

11:42

it's a workspace or

11:44

a religious space, the less

11:46

connection there is. So

11:48

what I'm wondering, you've been

11:50

looking into this, is

11:52

there going to be a

11:54

day maybe not too

11:56

far off in the future,

11:58

where we're really going

12:00

to be universes of one?

12:02

It's probably one of

12:04

the questions that I've been

12:06

thinking about the most

12:08

over the. the last 10 years. And it's

12:10

an interesting problem, because so one of the reasons

12:12

for why this happens, right? Like we talked about

12:14

how we generate a lot of data, we can

12:16

use that to get insights into who you are,

12:18

and most of the time these insights are used

12:20

to personalize your experience, right? So like Google uses

12:22

it to kind of give you the most accurate

12:24

or the most relevant search results, your news feed

12:26

on social media is personalized to what they think.

12:28

keeps you on the platform. And typically that

12:31

means that we cater to your

12:33

existing preferences. Right, so there's something that

12:35

we in psychology call exploitation versus exploration trade-off,

12:37

and it's in a way the way that

12:39

humans learn, right? So oftentimes we have this,

12:42

this almost like a tension between do we

12:44

go for the stuff that we know is

12:46

good for us, the stuff that we know

12:48

is good for us, the stuff that we

12:51

like. Think of going to a restaurant, right?

12:53

So there's like two ways in which you

12:55

can pick restaurants, either you go to one

12:57

that you know you love, because you love,

13:00

because every time it's good, is the expiration

13:02

part and that allows us to get better

13:04

because maybe the restaurant that we know is

13:06

great. If we explore and we try something

13:08

new there's always a risk that we stumble

13:11

upon a restaurant that just sucks right and

13:13

then we've missed an opportunity to just go to

13:15

the one that is tried and tested and we

13:17

like but also if we don't sample something new

13:20

we're never going to find the one that's kind

13:22

of blowing our mind and it's even better than

13:24

the one that we knew and I think that

13:26

is kind of missing. in the online space. And

13:28

there's reasons for that, right? Because like I was

13:31

just thinking when you said it would take you

13:33

so long to count to a million, I at

13:35

some point calculated something similar, like how long would

13:37

it take you to watch all of the content

13:40

on YouTube? And I think when if you assume

13:42

that the average person lives to about 90,

13:44

it would still be more than 2,000 years.

13:46

So the reason for why we stick to

13:48

this notion of I try to figure out

13:50

what you want and then I make recommendations.

13:52

is partially because there's no way that we

13:54

consume everything online. So it used to be

13:57

the case offline that there was a limited

13:59

number of comments. that we can join

14:01

limited number instead of news like

14:03

news outlets products that we can

14:05

buy now the world is your

14:07

oyster but there needs to be

14:10

something that helps us find stuff

14:12

that at least somewhat relevant but

14:14

what we've entirely lost in that

14:16

part worries me a lot on

14:18

many dimensions is we've lost this

14:21

exploration part. So we, like the

14:23

one thing that I find is

14:25

like nobody's really talking about could

14:27

make us so boring, right? If

14:29

we constantly see the same stuff

14:32

again and again, just based on

14:34

what we've done in the past,

14:36

we're never gonna stumble upon this

14:38

amazing restaurant that we didn't even

14:40

know was there. Simple algorithms, take

14:43

Google Maps, right? Google Maps is

14:45

like, you can think of it

14:47

as like a very simple AI.

14:49

It's kind of trying to figure

14:51

out how do you get from

14:54

A to get from A to

14:56

be. super helpful most of the

14:58

time, but we missed these opportunities

15:00

where we could just lost in

15:02

the streets and then you find

15:04

like this additional additional amazing coffee

15:07

shop. And I think there's a

15:09

way in which you could actually

15:11

fix this online. So like the

15:13

boring part is just one thing,

15:15

right? And the other one that

15:18

you mentioned that I'm equally worried

15:20

about is the splintering of what

15:22

I call shared reality. So that's

15:24

the idea that we can make

15:26

sense of the world together. And

15:29

it's a... fundamental, right? So why

15:31

is there such a big load

15:33

unless I put... epidemic. Why is

15:35

it that there's like a lot

15:37

more mental health issues? Partially I

15:40

think because it's breaking down and

15:42

I think there's ways in which

15:44

technology can actually help us solve

15:46

this. There's also much darker futures

15:48

that I see. I'm happy to

15:51

go into that but I'll stop

15:53

there to give you a chance

15:55

to wait. Thanks. I was processing

15:57

and I seized on the word

15:59

boredom. for a reason. Because one

16:02

of the things I've been thinking

16:04

about is that what these companies

16:06

have done in order to keep

16:08

people on their site. I mean,

16:11

they're kind of like a store

16:13

that has a customer walk in.

16:15

They have all these products and

16:17

they don't want the customer to

16:20

walk out the door and go

16:22

to the next door. And that's

16:24

perfectly understandable.

16:27

And so it's constantly,

16:30

because it understands

16:32

your choices from

16:34

the past, putting out

16:36

things that it knows

16:39

you're going to be

16:41

interested in, maybe even

16:43

fascinated in. Now,

16:45

when you think about that,

16:47

and what that is doing

16:50

to the minds of people

16:52

who are, say, 18 years

16:54

old, who never... had a

16:56

life beyond this technology is

16:59

it is eliminating boredom. I

17:01

mean, when I was a kid there were

17:03

times where like nothing was happening

17:06

on the street with your

17:08

friends and that was the

17:10

idea. What are we going

17:12

to do to make it

17:14

interesting? That's what I'm going

17:16

to say. It's just creativity.

17:18

Yeah. But now you're in a situation

17:21

where you don't ever have to

17:23

be bored. And not only

17:25

that, you may think it's

17:28

your right to never be

17:30

bored. That it's absurd

17:32

to be bored. And I don't

17:34

know where that's going,

17:37

but it's another case,

17:39

if you look at the image

17:41

you gave about not

17:43

following Google directions

17:45

and driving around,

17:48

looking for something,

17:50

stopping to see

17:52

somebody who's cutting their

17:54

lawn and saying, hey, you know, you

17:56

know where this restaurant is? And then

17:58

they say, oh yeah. I go there

18:00

all the time and now you've

18:02

got a conversation going. They know

18:04

the owner. Oh yeah, tell him

18:06

Frank sent you and like you're

18:08

in a different place. And I

18:10

know from talking to people that

18:12

there are like teenagers who have

18:14

trepidation, if not fear of actually

18:16

like calling up a pizzeria and

18:18

talking to somebody to order a

18:21

pizza. Now, they can easily send

18:23

the text. That's fine. But to

18:25

actually have to talk to somebody

18:27

they don't know or haven't sort

18:29

of been introduced to over the

18:31

internet is something that can maybe

18:33

even terrify young people. And you

18:35

tell us, where is that going

18:37

to take us? It's so funny,

18:39

because first of all, that's existed

18:41

always. I used to hate, I

18:43

could have used to pay when

18:45

I was younger, used to pay

18:47

my two years younger sister to

18:49

call the doctor, call the pizzeria,

18:51

because I just absolutely hated taking

18:53

phone calls. So I learned gradually.

18:55

So I do think that there's

18:57

a way even for the really

18:59

introverted people to get there. But

19:01

I think it's totally right. And

19:03

is that it just makes it

19:05

so much more convenient to converse.

19:07

in a way that is not

19:09

face to face. And it also

19:11

means like we're talking earlier about

19:13

AI just being there to cater

19:15

to exactly the way that you

19:17

want to be talked to, right,

19:19

tries to avoid conflict, always says

19:21

yes, because it's like meant to

19:23

please, right? The way that these

19:25

large language models work. they're trained

19:27

to be nice. We at some

19:29

point in the beginning, I think

19:31

a lot of people were trying

19:33

to do work on negotiations and

19:35

see if we could teach them

19:37

to negotiate and they will always

19:39

wait too nice because those are

19:41

the guardrails that the companies put

19:43

in place. But it also means

19:45

if that's the only counterpart that

19:47

you talk to, you're never going

19:49

to learn how to deal with

19:51

conflict. So I do think that

19:53

there's something that we think is

19:55

like a musled atrophy. Right. The

19:57

same way that because we use

19:59

Google Maps all the time, we

20:01

now have a hard time navigating

20:04

without. It's the same way with

20:06

conversation. If we don't talk to

20:08

people who think very differently, if

20:10

we don't talk to people who

20:12

push back and say, hey, that's

20:14

a stupid idea, and how are

20:16

we ever going to learn to

20:18

do this in real world relationships?

20:20

I do think that's true. What

20:22

I, so I'm going to give

20:24

you my optimistic view, and then

20:26

we can also talk about the

20:28

more pessimisticimistic one. we could actually

20:30

just repurpose AI in a way that

20:32

it does its exact opposite. So

20:34

AI is exhausting. AI doesn't really care

20:37

of do I cater to your preferences,

20:39

do I make your view narrow, do

20:41

I make you more boring? That's just

20:43

what the incentives are. But as you

20:46

said, the incentives of platforms are to

20:48

keep you there for as long as

20:50

possible. But AI in a way could

20:52

be the best perspective taking machine the

20:55

best. I think of it is like

20:57

an echo chamber swap. tool that we've

20:59

ever seen. So if I, for example, I

21:01

was using myself as an example, if

21:03

I wanted to know what is the

21:05

experience of a 50-year-old Republican farmer somewhere

21:08

in Ohio look like, I have no

21:10

idea. But I don't know what the day-to-day

21:12

experience looks like, and it would be

21:15

very difficult for me to find out.

21:17

Same for anybody else, because you'd have

21:19

to find multiple people, kind of... go

21:21

live their life for like a little

21:24

bit, exchange ideas. Now Google and Facebook

21:26

know exactly what it looks like, right?

21:28

Because they have their algorithm, they know

21:31

exactly what they see, they know what

21:33

their day to day look like, they

21:35

don't know what their news consumption look

21:38

like, they don't see what their news

21:40

consumption look like, they see what their

21:42

friends talk about, and they could just

21:44

instead of just optimizing for me and

21:46

keeping me in my little echo chamber,

21:49

see what they see. Now there's a

21:51

reason I think for why that doesn't

21:53

exist potentially is maybe we're not

21:55

going to use that all too

21:57

often, but it's super super comfy.

21:59

in the echo chambers that we've

22:01

built for ourselves, because they speak

22:03

to our preferences. It's like same

22:05

as an offline context. We typically

22:08

have friends who are very similar,

22:10

right? But once in a while,

22:12

we could actually break out and

22:14

I could say, maybe I'm even

22:16

going to get, the way that

22:18

I think about is like, I

22:20

would love to have an AI

22:22

guide, go to this echo chamber

22:24

with me, and try to help

22:26

me understand, right? Because if I

22:28

go there by myself, maybe it's

22:30

just... take me back right away,

22:32

but I could have someone who

22:34

understands it a little bit better

22:36

and who could help me bridge

22:38

their reality to mine and see

22:40

maybe here are some commonalities, maybe

22:42

here's why they think about this

22:44

specific topic in a certain way

22:46

that you might not think about.

22:48

I think that could be an

22:50

opportunity that we've never had, like

22:53

super expensive to travel, super expensive

22:55

to go to different parts of

22:57

the world, talk to the locals,

22:59

but that's a way in which

23:01

we could actually do that. So

23:03

that's my positive. What if we

23:05

could use it slightly differently? That's

23:07

your dream. That's my dream. It's

23:09

not that difficult because it exists.

23:11

So all of these models, they

23:13

already exist. All it would take

23:15

is for Google and Facebook to

23:17

say we're just going to have

23:19

an explorer mode where we repurpose

23:21

and we're going to say okay

23:23

instead of using your model, we're

23:25

going to turn your model off

23:27

for a second and we're going

23:29

to turn Cal's model on and

23:31

that's what you're going to see

23:33

for the rest of the rest

23:35

of your day. what you want

23:38

to see and when you want

23:40

to see it, would be easy

23:42

to implement. Question is, would companies

23:44

do it? I think that the

23:46

thing that I wonder about is

23:48

the speed that this has overtaken

23:50

us. And the same thing happened

23:52

with the internet, where everything seemed

23:54

like Disneyland. Hey, look at you,

23:56

Facebook, I talked to my friends,

23:58

Instagram, and then 10 years. later

24:00

you see that kids have been

24:02

shamed and bullied and in

24:05

ways that it's impossible for

24:07

their parents to comprehend or

24:10

or help them that they're

24:12

in in this world that's

24:14

just sucking second them down

24:17

you're talking about mental illness

24:19

and then it's like 10

24:21

years later oh let's do

24:23

something about this. Do we

24:26

have people who while we

24:28

may be behind or out there,

24:30

just trying to understand

24:32

where this is going,

24:35

it's sort of like the

24:37

Iroquois Indians used to have

24:39

a saying about, you know,

24:41

before you do something new,

24:44

you think about its impact

24:46

seven generations down the

24:48

road. I wonder if

24:51

anybody's thinking a day

24:53

beyond tomorrow. So here's my

24:55

personal answer. I think the one

24:57

thing that changed everything for me

24:59

personally is becoming a parent. I think

25:01

I was like very much focused on

25:03

what can we do right here and

25:05

now, what are the opportunities challenges, but

25:08

it was very limited. I think becoming

25:10

a parent suddenly kind of gave me

25:12

a much longer horizon. So this is

25:15

just like obviously a tiny, tiny microcosmas,

25:17

but I know that it changed something

25:19

for me. And the one thing that I'm thinking

25:21

a lot about in terms of. where that

25:24

potentially take us in the future

25:26

and it's I think it's mostly

25:28

dark with with some opportunities and

25:30

is this growing push towards

25:33

not just AI versus human right

25:35

so right now it's still where

25:37

human we have certain kind of

25:39

relationships with our friends family

25:41

the world and then there's AI

25:44

and yet there's like becoming we're

25:46

becoming closer and we're using them

25:49

as relational agents But there's still

25:51

some separation. I think there's a

25:53

growing push with lots of money

25:56

behind, mostly coming from Elon Musk,

25:58

to essentially merge the two. So the idea

26:00

that instead of having AI as an external

26:02

agent, can we actually put a chip in

26:05

our brain and now again it becomes a

26:07

little bit dystopian but bear with me because

26:09

I do think that there's some bearing in

26:11

reality is can we put a chip in

26:13

our brain so that instead of you having

26:15

to go to Google and say hey where's

26:18

the restaurant you just think the thought kind

26:20

of goes to the cloud and back. Now.

26:22

The reason for why I'm interested in that

26:24

space is because my husband is a neuroscientist

26:26

and we know, first of all, we know

26:28

how to read and write into the brain,

26:31

right? So we know how the language of

26:33

the brain works, so we know how to

26:35

read from the brain, and we also know

26:37

how to speak to the brain. So what

26:39

is happening at NeuralLink, for example, that's Elon

26:41

Musk Company, is that they're trying to build

26:44

chips that right now solve health issues, right?

26:46

like paraplegics, do kind of regain ability. Sounds

26:48

amazing, but the moment that you have people

26:50

walking around with chips in their brain, there's

26:52

so much more that you can do. And

26:54

so for me, this very dystopian far future

26:57

is like, well, if this world happens, first

26:59

of all, talk about this connection. Right, so

27:01

now suddenly we can get everything just by

27:03

thinking, I don't need to speak to another

27:05

human being, just face to face. Maybe our

27:08

chips can communicate at some point, but I

27:10

can be totally separated from the world because

27:12

everything just happens inside my brain. There's also

27:14

like, if we think about these echo chambers

27:16

from social media, you just broadcast to my

27:18

brain directly. And obviously, the moment that there

27:21

are some people who have that chip in

27:23

their brain and others don't, like inequality will

27:25

just widen. way way bigger than anything that

27:27

we've seen. So there would be people who

27:29

would talk about us as like these cute

27:31

little monkeys pressing buttons on their on their

27:34

computers while they just compute everything in their

27:36

head. So if we kind of really talk

27:38

about where this AI and all of this

27:40

stuff on terms of trying to figure out

27:42

who we are trying to change our behavior.

27:44

your leaders in the

27:47

future. I think that's

27:49

what I'm most worried

27:51

about right now. Again,

27:53

a pretty big leap,

27:55

but if you talk

27:57

about what should we

28:00

be talking about? Because

28:02

right now it's mostly

28:04

a legal question, right?

28:06

So unless you have

28:08

something like epilepsy that

28:10

requires you to have

28:13

a chip in your

28:15

brain that's already regulating

28:17

the way that your

28:19

body works, you're

28:22

not allowed to do it, but that's just a

28:24

legal barrier. Nobody says that once we see

28:26

the benefits in kind of curing some of the

28:28

disease, that we don't just kind of take

28:30

the step and say, okay, as long as it's

28:32

not too invasive, maybe we can put something

28:34

in. And then it's the same way that we've

28:36

seen it with the internet, like a pretty

28:38

slippery slope. So if I want to have people

28:40

talk about a conversation early, that

28:42

would probably be the one. Well,

28:45

and you haven't even brought up the fact

28:47

that what happens if the chip gets

28:49

hacked and you're told

28:51

to do something evil. Or

28:54

you don't even know if the thoughts that

28:56

you have are your own, right? So right now,

28:58

like you can at least note that something

29:00

that kind of pops up in your mind, maybe

29:02

it's been manipulated and maybe people have been

29:04

pulling the strings in terms of showing you different

29:06

new sources, but you at least know that

29:08

you're thinking the fun. Like once you have a

29:10

chip in your brain, this kind

29:12

of security has gone

29:14

entirely. So there's a pretty

29:16

big challenge associated with that.

29:18

Yeah. But it is getting

29:20

scarier and scarier. I told

29:23

you, this is the dystopian version of all

29:25

of this. Okay. So

29:27

there's, let's look at the other

29:29

side here. Let's

29:31

try to go back. In two ways. In two

29:33

ways. What can

29:36

we do now

29:38

to just try to

29:40

protect ourselves from just

29:42

giving away too

29:44

much of ourselves and

29:47

maybe even starting

29:49

with understanding how

29:52

pieces of ourself

29:55

are being turned into data that

29:57

we're not even aware of.

30:00

How much time would that take? Because

30:02

it might be such a huge

30:04

amount of time that most people

30:06

would just throw up their hands

30:08

and say, look, I don't care.

30:10

I'm just going to get my

30:12

coffee. If six gigabytes of data

30:14

goes out because I put my

30:16

credit card through it, I

30:19

really don't care. And the younger

30:21

the people are, probably the easier

30:23

it is for them to do it

30:25

because that's their world. I think it's

30:27

a great question and I again like

30:30

we were trying to pivot back to

30:32

the to the bright side I think

30:34

I've become a lot more pessimistic about

30:36

this space over the last couple of

30:38

years I would say and so I

30:40

do think that we can all do

30:42

a little bit better right so I

30:44

think first of all most of us

30:46

when we think about our data being

30:48

very intrusive and privacy being violated. We

30:50

mostly think about social media, which makes

30:52

sense because that's what's constantly being shown

30:55

in the media. That's what everybody discusses

30:57

in the public space. But it's such

30:59

a tiny, tiny slipper. So some people

31:01

say, well, I don't want to use

31:03

social media because I don't want to

31:05

be tracked. But then they download the

31:07

weather app and they kind of mindlessly

31:09

say yes to having the app tab

31:11

into their microphone, their entire photo gallery,

31:13

their GPS records. which is just as

31:15

intimate, right? So I think sometimes, especially

31:17

with phones, that's just like the one

31:19

thing that I would recommend kind of

31:21

people to be a little bit more

31:23

mindful because you do have a bit

31:25

more control there and then in other

31:27

parts of your life. However, the one

31:29

thing that I've become more pessimistic is

31:31

that if we're put in charge of managing

31:34

our personal data, it's never going to

31:36

happen. Right. So even if you look

31:38

to some of the regulations that are

31:40

out there now, even that the most

31:42

progressive ones. The most common path to

31:44

trying to solve it is to say, well,

31:46

why don't we just tell people what's happening

31:48

with their data so we mandate transparency? And

31:51

then we give them control so they can

31:53

take care of themselves. It's just never

31:55

going to happen because, like, first of

31:57

all, technology developed so fast, right? You

31:59

talk. about like the rapid exponential

32:01

speed by which technology grows. So

32:03

I do this as a full-time

32:05

job and I think about this

32:07

all the time and I can

32:09

hardly keep up with the development

32:11

in technology. And then even if

32:14

people just magically caught up, right,

32:16

that say everybody gets educated early

32:18

on in schools and they just

32:20

keep up, it's a full-time job.

32:22

If you really wanted to make

32:24

sure that you read all of

32:26

the terms and conditions and you

32:28

kind of manage all of the

32:30

permissions. that would be like a

32:32

24-7 and most people coming back

32:34

to being boring hopefully have better

32:36

stuff to do than just reading

32:38

through all of the terms I

32:40

really hope so right so I

32:42

think that the solution where we

32:44

just say well people need to

32:46

understand and then they need to

32:48

better manage is never going to

32:50

happen the brain is not made

32:53

for these choices right if the

32:55

brain can say oh I'm going

32:57

to use the service right here

32:59

now by clicking yes or you

33:01

can spend the next two hours

33:03

trying to decide for everything, we're

33:05

absolutely going to click gas. So

33:07

we kind of, what we need

33:09

is to make it much easier

33:11

for people to do the right

33:13

thing. And some of that I

33:15

think is regulation, right? So not

33:17

only sends a signal of what

33:19

we care about as society, but

33:21

it also can say, well, why

33:23

don't we make the default that

33:25

your data is not being tracked?

33:27

But now nobody does that, right?

33:29

We're lazy. So instead of making

33:32

laziness work against us, we could

33:34

actually make it work in our

33:36

favor. You know, that's kind of

33:38

interesting. And you're the one who

33:40

brought up the swiping of a

33:42

credit card. Now, if you want

33:44

to use a credit card, then

33:46

you have to apply and then

33:48

you're back in that same place,

33:50

you know, agree to these conditions.

33:52

You want this, you give us

33:54

that. Yeah. And it doesn't seem

33:56

like the text side is going

33:58

to take away any. that can

34:00

help them understand me

34:02

better so that it could either

34:05

keep me on its platform

34:07

or get me to buy something at

34:09

selling. Yeah, there's actually,

34:11

so I think you're

34:14

absolutely right because right now it's

34:16

a binary choice. Right. So the only

34:18

choice that we have right now is

34:20

give us all of your data or

34:22

don't get. the superior service or the

34:24

product, use the product at all. And

34:26

I think if that's the choice, the

34:28

brain is always going to go for,

34:30

no, I do want the better service

34:32

in the here and now. But I

34:34

do think that there are solutions

34:37

that we're not talking about enough and

34:39

for a reason because there's many businesses

34:42

that are just in the, they're making

34:44

money from. commercializing your data.

34:46

But there's many companies that are not.

34:48

And so there's this one thing that

34:50

I think everybody should know about. And

34:52

I'll try to explain it in non-technical

34:55

terms, but it's called, for those who

34:57

want to know, it's called federated learning.

34:59

And it's a way of creating

35:01

intelligence and machines and predictive algorithms without

35:03

collecting your data. So I'm going to

35:06

give you an example to just make

35:08

it a bit more concrete. So

35:10

let's take Netflix. Like we all. Watch

35:12

movies on Netflix and in a way it's

35:14

helpful that they help us find the movies

35:16

right because there's so many movies out there

35:18

You'd never be able to find what you

35:21

want if it was just up to you

35:23

The way that it typically works like the

35:25

way that typically machines are being trained is

35:27

you have your data like the stuff that

35:29

you've got the movies that you've been watching

35:31

the ratings you submit that to Netflix all

35:34

of the data gets sent and now it

35:36

sits there Netflix on Netflix on a server

35:38

and they train so they know that well

35:40

Cal really liked Titanic and he also really

35:42

liked Love Actually. So here is like

35:44

the recommendations that we're going to make

35:47

for him and here's how we're going

35:49

to update the model. Now this means

35:51

that you had to send your your

35:53

data to Netflix, but what we can

35:55

do right now is instead of you sending

35:57

all of your data, they can essentially

36:00

send the intelligence to you, right?

36:02

So your phone is so much

36:04

more powerful that again, the rockets

36:06

that we use to send people

36:08

into space with. So they don't

36:10

need to get all of the

36:12

data. They can send the model,

36:14

the intelligence to you and say,

36:16

okay, locally, we see that you've

36:18

watched Itonic, you've watched lab, actually,

36:20

that never gets shared with Netflix

36:22

at all. The model just learns

36:24

and updates its little parameters, learns

36:27

a little bit about how the.

36:29

the movie world works, and instead

36:31

of you sending the data, you

36:33

just send back some of the

36:35

intelligence. So now the model of

36:37

Netflix gets better, we all benefit,

36:39

but you've never had to send

36:41

your data. And Netflix is just

36:43

like an example that might not

36:45

seem like, well, why would I

36:47

care sending my data to Netflix?

36:49

But think about medical data. Right,

36:51

so like there's so much that

36:54

we could do and learn about

36:56

diseases, how do they work, what

36:58

kind of treatments work for which

37:00

type of person under which conditions

37:02

and so on. Right, right, right,

37:04

right. All of this genetic data,

37:06

medical histories, biometric data, but I

37:08

don't want to farm a company

37:10

to have all of this data

37:12

centrally, right, because like now that's

37:14

a huge risk of, first of

37:16

all, data breaches, then abusing it

37:19

now, maybe they get a new

37:21

CEO, that is like once my

37:23

data is out there I can't

37:25

get it back but what they

37:27

could do is they could say

37:29

I'm going to keep my genetic

37:31

data you're going to keep your

37:33

genetic data they just kind of

37:35

send questions to my data so

37:37

they figure out okay over time

37:39

here's how like certain social demographics

37:41

respond to this kind of treatment

37:43

and maybe this works better for

37:46

women and so the data stays

37:48

with us and we're just sharing

37:50

back some of the intelligence And

37:52

it's a totally different way of

37:54

learning. And it really breaks down

37:56

this dichotomy that we've, I think

37:58

Silicon Valley has very carefully crafted

38:00

for a long time, where it's

38:02

like, well, either you have to

38:04

give us your data, because that's

38:06

the only way that you can

38:08

give the service and convenience and

38:11

like better medication and better product,

38:13

but that's not. longer true and

38:15

I think once we move to

38:17

a model that's more decentralized we

38:19

have a lot more power because

38:21

the data stays with us and

38:23

we have a lot more control

38:25

over over how it's being used.

38:27

So I think that's a model

38:29

that again makes it a little

38:31

bit easier to keep control and

38:33

exercise it more wisely. I don't

38:35

know that many companies would want

38:38

to do that because they already

38:40

have the upper hand and then

38:42

to it's asking them to do

38:44

extra work. Although, with the speed

38:46

of AI now, that may not

38:48

be that big of a problem.

38:50

But let's take it to pivoted

38:52

toward solutions and ways that we

38:54

could make things better. You thought

38:56

a lot about this. And actually,

38:58

before I even asked you to

39:00

get to the solutions, you were

39:03

just saying how much time it

39:05

takes to just try and keep

39:07

up with it. You didn't even

39:09

say trying to stay ahead. You

39:11

said just trying to keep up.

39:13

Is there any formula you could

39:15

recommend to somebody like me who

39:17

also wants to keep up? How

39:19

do you keep up? It's incredibly

39:21

difficult. So I think I'm in

39:23

right now a relatively lucky position

39:25

is that oftentimes people send me.

39:27

stuff so like I work a

39:30

lot with students and I think

39:32

by now people know what I'm

39:34

interested in so a lot of

39:36

the cutting-edge stuff actually comes from

39:38

other people recommending. I personally like

39:40

Wired because I think their articles

39:42

are really interesting I think they

39:44

have a very nuanced take on

39:46

what are some of the opportunities

39:48

and also some of the challenges

39:50

so and then also for us

39:52

again it's a little bit easier

39:55

because we have like all of

39:57

these academic institutions but I do

39:59

think that longer forms like wired

40:01

or podcast I think go a

40:03

long way, but again, it's a

40:05

it's an uphill battle if you

40:07

want to keep informed on that

40:09

also like times actually has an

40:11

interesting section on data and privacy

40:13

that is that is really good.

40:15

But again, those are just who

40:17

I'm sure there's many more. Okay.

40:19

So what are your best hopes

40:22

for pushing this forward in a

40:24

way that's positive for us and

40:26

a way that we're not going

40:28

to lose ourselves? I mean, I'm

40:30

already now thinking about what's going

40:32

to happen. when chips start going

40:34

into people's heads. And you know,

40:36

this is science fiction stuff. Do

40:38

you have any real world solutions

40:40

to pass on to us? Yeah,

40:42

there's actually, there's some hope. And

40:44

actually, I'm quickly going to go

40:47

back to something that we talked

40:49

about before, because one of the

40:51

things that I'm trying to do

40:53

is actually convince companies that it's

40:55

in their best interest, because I

40:57

don't think if we don't have

40:59

the corporate sector on board. there's

41:01

only so much that we can

41:03

do right so I think regulation

41:05

is slow it's super difficult to

41:07

police and implement so I actually

41:09

do think that there's benefits for

41:11

for companies because unless you're in

41:14

the business of selling data right

41:16

so Facebook for example is as

41:18

much as it they kind of

41:20

want to be a B to

41:22

see and focus on the customer

41:24

they just sell your data to

41:26

companies but if that's not the

41:28

case right say you're in the

41:30

space where you're trying to offer

41:32

better products, you're kind of trying

41:34

to make better recommendations. Ditting on

41:36

people's data is a huge risk.

41:39

So it used to be the

41:41

case that you needed to collect

41:43

people's data to offer your services,

41:45

right? If you want to make

41:47

recommendations, there was just no other

41:49

way than collecting the data centrally.

41:51

But if you think about it,

41:53

you've just accumulated this pile of

41:55

gold, you've just accumulated this pile

41:57

of gold. you're actually like now

41:59

you're in this place where you

42:01

have to protect it and like

42:03

the reputational costs legal costs of

42:06

someone breaching the data and even

42:08

protecting it having a tie in

42:10

a system is super expensive So

42:12

if you can offer the same

42:14

services, the same convenience, the same

42:16

product, without having to collect it

42:18

centrally, you're in a much better

42:20

spawn. Right? Not just that, but

42:22

then you can also make it

42:24

part of your value proposition. Because

42:26

if you can say, hey, we're

42:28

going to give you the same

42:31

experience without collecting your data, then

42:33

our competitor does. Yeah, you know

42:35

what? That's a great idea. Apple

42:37

has figured it out, right? Apple

42:39

has figured it very beat to

42:41

customer. And if you look at

42:43

in New York, the ads that

42:45

Apple has, it's all about privacy.

42:47

And they're actually one of the

42:49

big ones using federated learning. So

42:51

Siri, for example, on your iPhone,

42:53

it doesn't send the data to

42:55

the cloud. It kind of that

42:58

they sent a model to your

43:00

phone. It kind of that they

43:02

sent a model to your phone,

43:04

it locally, sent a model to

43:06

your phone, it kind of that

43:08

they sent a model to your

43:10

phone, to commercialize. They have no

43:12

incentive what I'm trying to do.

43:14

to make the world better, the

43:16

way that we talked about it.

43:18

And I do think it's, I'm

43:20

not just making this up, I

43:23

do think that there's a real

43:25

incentive that companies have for consumers.

43:27

The way that I hope this

43:29

is going to go is essentially

43:31

trying to build these new forms

43:33

of data governance that just gives

43:35

us a lot more support. So

43:37

we talked about. Well there's amazing

43:39

ways in which we could benefit

43:41

from biomedical data but we just

43:43

don't have the time to think

43:45

about all of the implications, who

43:47

does we share the data with,

43:50

for what purpose, because again we

43:52

have like only 24-7. There's a

43:54

really intriguing idea that's called data

43:56

co-ops and data trusts, that it's

43:58

like essentially a way in which

44:00

people can come together with a

44:02

shared interest in data. My favorite

44:04

go-to example these days is, and

44:06

you can see a common theme

44:08

here, is expecting moms. So when

44:10

you're pregnant, right, so you have

44:12

so many questions, you want to

44:14

know what are my risk factors,

44:17

what are the baby's risk factors,

44:19

what should I be doing? And

44:21

what you get is like every

44:23

four weeks you get to see

44:25

your... doctor and they say yeah

44:27

it's still breathing and the heart

44:29

is still beating and it's just

44:31

like terrifying now we could have

44:33

like an amazing kind of service

44:35

that says well let's have like

44:37

a lot of pregnant moms come

44:39

together pull our genetic data again

44:42

medical histories biometric data lifestyle choices

44:44

so we learn Like what should

44:46

each woman do given the back

44:48

genetic background and so on to

44:50

make sure that she's safe and

44:52

the baby is safe every step

44:54

along the way? And I don't

44:56

want to farm a company to

44:58

have that data, but if we

45:00

can come together and create a

45:02

data cop or data trust, we

45:04

can actually hire management so we

45:06

can hire people with expertise in

45:09

the use of data, they can

45:11

help us exactly figure out whom

45:13

should we be sharing with, they

45:15

can help a set of the

45:17

infrastructure for federated learning. Right, so

45:19

now we have like many people

45:21

with a shared interest, they can

45:23

make sure that we have the

45:25

technological infrastructure and the same way

45:27

that banks have a fiduciary responsibility

45:29

to act in the best interest

45:31

of their customers, that management is

45:34

now legally obligated to act in

45:36

our own best interest. And so

45:38

this is like a totally different

45:40

way of thinking about how do

45:42

we not just kind of eliminate

45:44

some of the risks through regulation,

45:46

but how do we actually help

45:48

us optimize and maximize the value

45:50

that data could have for all

45:52

of us? without us having to

45:54

be solely responsible for doing it

45:56

all by ourselves. So that's for

45:58

me, one of the most interesting

46:01

solutions. Now are you just putting

46:03

this out as an idea or

46:05

are you trying to actually implement

46:07

it somehow? Because as you're talking,

46:09

I'm thinking, wow, there are a

46:11

lot of diseases that aren't particularly

46:13

common. Yeah, that yet like a

46:15

doctor might see it may be.

46:17

once or twice a year. And

46:19

so there's very little experience that

46:21

the doctor can have, but if

46:23

you had all the data from

46:26

around the country or around the

46:28

world. Yeah, it already exists and

46:30

George hitting. the head on the

46:32

nail. So essentially my favorite example

46:34

of that is called my data

46:36

and my data is a Swiss

46:38

data cop in the medical space

46:40

and they focus exactly on what

46:42

you were just describing. So one

46:44

of the applications that they have

46:46

looks at MS, so multiple sclerosis,

46:48

which is one of the disease

46:50

that's absolutely crippling and we don't

46:53

understand it. Right. So it has

46:55

like multiple determinants. It's like genetic

46:57

but it's also kind of environmental

46:59

and so there's many, many ways

47:01

in which we could understand the

47:03

better. So my data is like

47:05

the Swiss data co-op that gets

47:07

a lot of data from patients

47:09

and again they own the data

47:11

co-op. So they have like not

47:13

just control over their data, they

47:15

also have stay in what the

47:18

data co-op does through like this

47:20

general assembly. And it's amazing because

47:22

once you collect data from many

47:24

MS patients and also like the

47:26

community of people who doesn't have

47:28

MS, right? Because you need a

47:30

comparison. And first of all you

47:32

can scientifically understand disease much better.

47:34

But what they do in addition,

47:36

which I think is like just

47:38

extraordinary, is for every patient, they

47:40

now can have a predictive model

47:42

and it connects with their doctors.

47:45

So it's not just an AI,

47:47

it's essentially an AI saying, oh,

47:49

here's something that we've learned about

47:51

this patient from this massive data

47:53

set. It sends this intelligence to

47:55

their local doctor. They kind of

47:57

look at how does it interact

47:59

with the medication and then the

48:01

doctor gives feedback to. the algorithm.

48:03

So it says, okay, this medication

48:05

actually works for this patient. I'm

48:07

seeing improvement in the way that

48:10

they kind of interact. And this

48:12

to me is like, just again,

48:14

mind blowing because now you have

48:16

like scientific knowledge that is being

48:18

built in a way that we

48:20

would have never been able to,

48:22

right? Maybe farm companies, if they

48:24

think there's profit, and then maybe

48:26

you never get to benefit from

48:28

it because it only happens in

48:30

20 years. Best case, you have

48:32

to pay millions of dollars to

48:34

get the thing. And again, the

48:37

data corp just acts in your

48:39

best interest and it helps you

48:41

benefit in the here and now.

48:43

So it's not hypothetical and already

48:45

exists and I think that there's

48:47

probably going to be a lot

48:49

more of those. I like where

48:51

you're going. It sounds like it

48:53

exists, but it takes a lot

48:55

of work. And it almost seems

48:57

like it's got to be somehow

48:59

forced or pushed the same way

49:02

you were talking about Apple and

49:04

their ads about privacy. Like, you

49:06

know, once you have a push

49:08

in that direction, the competitors have

49:10

to compete against it. So, you

49:12

know, it seems like. the first

49:14

of many conversations I'd like to

49:16

have with you. I don't know

49:18

if you're open to some more.

49:20

I would love to. Are you

49:22

easy? I could talk about this

49:24

forever. I'm sure you got a

49:26

sense. Well, I did. And I

49:29

could ask questions about it forever.

49:31

So this is a good match.

49:33

Let me thank you for now.

49:35

And let me just watch where

49:37

this world is going. You watch

49:39

where the world is going. And

49:41

then I'm sure down the road.

49:43

Down the road. our paths are

49:45

going to meet again because your

49:47

book is fantastic. It's a rare

49:49

business book that is written with

49:51

like a storytelling vitality from first-person

49:54

point of view that really puts

49:56

us in your shoes. And when

49:58

you find somebody who can explain

50:00

these complicated topics on a very

50:02

personal storytelling level, The world is

50:04

a better place for it. So

50:06

thank you for joining me today.

50:08

Thank you so much for having

50:10

me. And about wraps it up.

50:12

Want to thank Tim Ferris for

50:14

nudging me to start this podcast.

50:16

As we step into 2025, concept

50:18

of trust is being redefined. Artificial

50:21

intelligence is the power. to transform

50:23

our lives, making us more productive,

50:25

even more creative and connected. never

50:27

before. Yet, it also brings challenges

50:29

we've never faced. Deep fakes, data

50:31

manipulation, technology capable of knowing us

50:33

better than we know ourselves, and

50:35

deceiving us at scale. I am

50:37

craft in a keynote speech that

50:39

explores a critical question. How do

50:41

we build trust in a world

50:43

where we can no longer believe

50:46

everything we see or hear? To

50:48

find answers, I've been reflecting on

50:50

conversations I've had with iconic leaders

50:52

who've shaped history. Their insights offered

50:54

timeless lessons about the pillars of

50:56

trust that have held us together.

50:58

At the same time, I'm diving

51:00

deep into the future, exploring how

51:02

AI is reshaping our relationships with

51:04

information and with one another. This

51:06

is the keynote for our times,

51:08

and it will empower audiences to

51:10

navigate the complexities of our rapidly

51:13

changing world. It's not just about

51:15

identifying the challenges, it's about uncovering

51:17

the opportunities to strengthen trust all

51:19

around us. Business, leadership, and society.

51:21

If trust is a priority for

51:23

you, or your organization in 2025,

51:25

let's talk. I'd be happy and

51:27

honored to deliver this keynote at

51:29

your next event. Do you have

51:31

any thoughts on trust in the

51:33

age of AI? I'd love to

51:35

hear them. Please, reach out to

51:38

me at Cal Busman.com. Here's to

51:40

a year, filled with connection, curiosity,

51:42

and trust. Cheers!