0:15

Pushkin. I'm

0:20

Maybe Higgins, and this is solvable

0:23

Interviews with the world's most innovative

0:26

thinkers who are working to solve the

0:28

world's biggest problems. My

0:30

solvable is that every frontline

0:33

social organization is the ability

0:35

to use data and AI same way,

0:37

same capacity that the big tech companies

0:40

do today. I want to see a world where the

0:42

same algorithms that are routing your packages

0:44

to you your house coming so efficiently

0:46

because an AI figured out the best way to avoid traffic

0:48

and weather are just as equally being applied

0:51

to delivering a vaccine through an area

0:53

before it spoils. That's Jake

0:55

poor Away, the founder and CEO

0:58

of nonprofit Data Kind. He's

1:00

talking to Jacob Weisberg about how

1:02

he's working to make that world

1:05

a reality. The Rockefeller Foundation

1:07

has thought about this too. More than

1:09

two point five quintillion bytes

1:12

of data are produced every day.

1:14

That's one hundred trillion bytes.

1:17

This abundance of data, combined

1:19

with rapidly advancing analytics

1:21

capabilities, could really improve

1:23

the lives of billions of people

1:25

around the world, but it's

1:28

only living up to a fraction of that potential.

1:31

While private sector businesses have been building

1:34

and deploying data science capabilities

1:36

for many years now. Most organizations

1:39

in the nonprofit and civic and public

1:41

sectors are way behind. Of

1:44

course, they want to use the applied

1:46

data to make their work go farther and

1:48

faster and to help more people, but

1:51

they don't often have the resources.

1:54

I mean, put yourself in the shoes of a newly

1:57

minted graduate. They're probably

1:59

wearing tivas. Last year, the

2:01

San Francisco Chronicle analyze

2:03

glass door data of the starting salaries

2:06

of some of the biggest tech companies in

2:08

the Bay Area. They found

2:10

out that tech pays even for the

2:12

young and inexperienced. The

2:14

average starting salary for a software

2:16

engineer was almost ninety two thousand

2:19

dollars. So there's

2:21

the workers and then there's the technology

2:23

itself. We know the power data

2:25

science can have for social good because

2:28

we've seen it in action. When mission

2:30

driven organizations have the right talent

2:32

and tools and knowledge, data

2:34

science can generate real human impact,

2:37

helping vulnerable families access public

2:40

benefits, saving water and

2:42

money during droughts, and saving time

2:44

in resettling refugees so

2:46

that they can find homes and jobs

2:49

faster. Jake Borway

2:51

works on this stuff every day.

2:53

He's a machine learning and technology

2:56

enthusiast who loves nothing more than seeing

2:58

good values in data. In

3:00

twenty eleven, he found a data Kind,

3:02

bringing together leading data scientists

3:04

with high impact social organizations

3:07

to better collect, analyze, and

3:09

visualized data in the service of humanity.

3:12

Jake works to ensure organizations

3:15

like the Red Cross have access to AI

3:18

and data science that's as good as

3:20

the access enjoyed by huge companies

3:22

like Facebook. Data Kind has

3:25

twenty thousand volunteers around the world,

3:27

who he likens to mets on San Frontier,

3:30

the doctors without Borders, except their

3:32

data scientists working pro bono

3:34

with leading social change organizations

3:37

on all kinds of projects,

3:39

including one that has data scientists

3:41

from Netflix predicting water usage

3:44

in a California neighborhood. It's

3:46

fascinating, So enjoy this conversation

3:49

and I'll talk to you after. What's

3:56

the problem. In a nutshell, the

3:58

problem is that digital technology

4:00

and artificial intelligence

4:02

have exploded over the last ten or fifteen

4:05

years, which have created huge opportunities

4:08

in the corporate space or

4:10

in building new apps for society, but

4:12

there's very little application

4:14

of that to social sector

4:16

causes. So we have this huge

4:18

opportunity to use a revolutionary technology

4:21

to predict the future

4:23

of things, to understand our society better,

4:25

to automate things that we either don't want to or couldn't

4:27

do, And yet there's a huge

4:30

potential loss in that it's very difficult

4:32

to get that applied to pro social causes

4:34

that we need. Jake is a data

4:36

scientist. When did you start

4:39

to see some of the downsides

4:41

around big data? Really? The article

4:44

that I used to point to is like the beginnings

4:46

of the tide turning to the negative. Was

4:49

the article that was titled

4:51

very salaciously, Target Knows

4:53

You're pregnant, And if you remember this one

4:55

from twenty thirteen, but the basic idea

4:57

was that someone had their

4:59

daughter, that maybe sixteen seventeen year old daughter

5:02

was receiving mailers from Target that said,

5:04

Hey, we think you need to buy kupons

5:06

for baby diapers or formula,

5:09

and the dad called up, you know Target, all, Matt,

5:11

So what are you sending me all my daughter all these

5:13

deals for having babies. She's not pregnant,

5:15

Like, why are you trying to get her to become pregnant?

5:18

And the person on the other end of the line, of course

5:20

didn't know what was happening, because you know, the

5:22

algorithms just send you what they think

5:24

you're going to buy based on other stuff you've bought, and

5:28

it's He called back later, kind of shame facedly and

5:30

said, you know, I talked to my daughter and actually she is

5:32

pregnant, and you know, the data had

5:34

picked up on that simply because you know, it watched what she

5:36

bought and she was probably buying you know, prenatal care,

5:39

vitamins and stuff. But that article

5:41

got shared around as the

5:44

sign that big data was going to

5:46

be negative. Target knows you're pregnant.

5:48

What a horrible invasion of privacy. That title alone

5:50

should, you know, make everyone's skin crawl.

5:53

But that's the problem is that that shouldn't

5:55

be the case. We think of there are so many

5:57

opportunities to be using data and

5:59

algorithms to see where

6:02

disease outbreaks are going to occur or predict

6:05

in the same way as what kind of conditions

6:07

you might have so you can live a healthier life.

6:09

And so I think it was then that we really thought,

6:12

Okay, we need to come out

6:14

and show the positive sides of this. Otherwise

6:16

everyone's going to just run to the fear around

6:19

what data science can do. We're

6:21

interested on this podcast and

6:23

people who've taken this leap to become

6:26

problem solvers and to take on the

6:28

biggest problems in the world. What

6:31

made you take a leap

6:33

to leave the private sector

6:36

to start an organization with

6:38

an ambitious goal. Well, I

6:40

have to say it was a bit of an accident. Actually it

6:43

was maybe twenty ten or eleven, and

6:45

I had just coincidentally come out

6:47

of school with a computer science and a statistics

6:49

degree, which little did I know was going to become

6:52

what would lead to the title data scientist. And

6:55

I was working at the New York Times R and D Lab,

6:57

and really what seemed obvious

7:00

was the fact that we had all of this new digital

7:02

technology, from cell phones that people

7:04

were carrying around with them, to satellites

7:07

launching in the air, to sends

7:09

being put around the world, that we were digitizing

7:11

our very existence. We were becoming

7:14

a digital species. There was almost like a central

7:16

nervous system to the world, and that

7:19

meant that were these huge opportunities to

7:21

learn from that to you know, have algorithms

7:24

drive maybe our greatest human values.

7:26

But the folks who really knew how to convert data

7:29

into those actions. The data scientists

7:31

were largely locked up in tech companies, and

7:35

you know, I would actually go to hackathons,

7:37

which are you know, like weekend events where technologists

7:40

would get together and just work on whatever

7:42

they thought was cool. And I would sit there

7:44

and think, this is so interesting because you know, we're not

7:46

at a company, we're not at our jobs.

7:48

We're here on the weekend. You know, I'm sitting next

7:50

to some machine learning engineer from Google and

7:52

NASA scientist, and I'm like, this is great. We

7:55

can make whatever we want. Like the world

7:57

has just become so ripe for what's possible.

8:00

And at the end of the day, the stuff that people made was

8:02

just so unfulfilling.

8:04

You know that someone had made like Twitter for pets,

8:07

or had improved how you'd find local

8:09

deals in your neighborhood, and

8:11

so I just said, man, there's got to be something

8:14

more we can do for society, or something

8:16

more fulfilling really than this, as opposed to

8:18

solving the problems of very well paid

8:20

twenty somethings in the Bay Area, right,

8:22

which is the parody, but that is a lot of

8:24

the new companies you hear about are

8:27

solving problems like how do you get your food

8:29

delivered or god knows how to get cannabis

8:31

delivered? You know when you when you could already

8:34

buy it by walking around the corner. You're exactly

8:36

right. We solve the problems that we ourselves

8:38

have. And as you've pointed out, the tech community

8:40

for better for worse, excused young male

8:43

US. So, yeah, I just thought, you

8:45

know, what would it take for to be applied to the social sector.

8:48

Where are the people who are on the front lines of getting

8:50

people food or clean water? And how could you apply

8:52

it there? And so I just wanted that

8:54

job myself. What didn't exist?

8:57

So I just wrote to a couple of folks in the

8:59

community here in New York and said, hey,

9:01

you know, instead of going and building you know,

9:04

a door dash competitor, could

9:06

we, I don't know, work with the Red Cross

9:09

US or Kiva who goes cash

9:11

transfers to folks, and say what could we do with

9:13

their data? What could we learn? What are the positive

9:15

ways we could work together with them? And

9:18

I thought people would just say, yeah, good idea, Jake,

9:20

but no thanks. I kind of just

9:22

buried the little sign up

9:24

link for folks, and I

9:26

was surprised to find that people started sharing around before

9:29

I knew it. I came back to work the next time,

9:31

hundreds of emails in my inbox from people

9:33

not just in the city but around the world saying, though

9:36

this is great, I want to get involved with data

9:38

kind, I want to do data kind France. At

9:41

one point, a few months into this, the White House

9:43

called and said, hey, we're interested in big data initiatives.

9:45

What's this thing? And you know, joke because

9:48

I don't know, it's not really

9:50

a thing, But to me it really tapped

9:52

into an energy from both

9:54

the technology side and the nonprofits

9:56

and governments who are writing, who said, we're

9:59

energetic to take on this new

10:01

wave of this technology and figure out how could

10:04

be applied. And so our job ever since

10:06

has really just been trying to support that

10:08

community, harness its energy, and be

10:10

helpful in any way we can. Since

10:12

you've been doing this, it's amazing how quickly

10:14

attitudes have shifted around

10:17

big data and algorithms. I mean,

10:19

just think about Facebook, which even

10:21

a few years ago was thought as

10:24

a socially positive

10:26

company. That was why part of why people went

10:28

to work there, and in just

10:30

a couple of years it's become

10:33

something that people think is an overwhelmingly

10:36

negative force. Are we're swinging too

10:38

far in the other direction in our skepticism

10:41

about what data is going

10:43

to be used for? Well, I think there's

10:46

a healthy reckoning on how

10:48

we've been using data and technology in the

10:50

past. You're right that in the last

10:52

couple of years there was sort of unfettered techno

10:54

optimism amongst a lot of the big

10:56

companies and that this would just change everything and

10:59

nothing could ever go wrong with social media

11:01

and data. So I think there is an obviously

11:04

very healthy reckoning of this, and we're starting to realize

11:06

what the downsides could be. What

11:09

your point I think is missing and we really need

11:11

to get acclimated to, is where

11:13

do we go from there? You know, is the

11:16

idea that we're just going to put the genie

11:18

back in the bottle, not use digital

11:20

information in these ways, regulate

11:22

all companies into existence. I'm

11:25

in favor of, by the way, stronger regulation,

11:27

for sure, But I think what we need

11:29

now is more examples and

11:31

more of a community of practice around what

11:34

it looks like to use these technologies ethically.

11:36

That's a big conversation obviously, that's in the space

11:38

right now. You hear a lot about the ethics of data

11:41

use, ethics of AI, but even then

11:43

I find those conversations fairly academic. I

11:45

think what we need are some more positive examples

11:47

of how it can be applied and positive principles

11:49

that we all agree to adhere to. And

11:51

so the data kind that's something we're

11:54

really working to try to demonstrate,

11:56

is to say, yes, we need

11:58

to protect ourselves, uphold

12:01

our civil liberties through data. Make

12:03

sure that we're not degrading human life

12:05

with what's going on with data in the business

12:07

world? And what does

12:09

it look like when you want to use

12:12

data and algorithms to predict,

12:14

say, inclement weather that could wipe out a

12:16

crop and that's critical to someone's

12:19

sustenance in another part of the world. What's

12:21

the good version of this? You know? How do you make

12:23

sure that it's accountable to those folks? How do we

12:26

make sure that everyone involved has

12:28

some sense of what the algorithm is doing and how their

12:30

data is being used. And I don't think

12:32

we can move past that point just by talking

12:34

about it. I think we need real concrete

12:37

examples of data scientists,

12:40

nonprofits, social organizations, constituents

12:42

getting together to say, what does the good version

12:44

of this look like a better version. I should say

12:46

there was a positive example in the news

12:48

recently with the prediction of

12:51

the cyclone in South

12:53

Asia that killed

12:55

very few people, and in the world before

12:58

big data, that same storm

13:00

might have killed a lot of people through panic,

13:03

through all sorts of consequences

13:05

because people wouldn't have known it was coming. I mean,

13:08

is that the kind of example we're talking about

13:10

here? Something positive? I think that's exactly

13:12

right. So at data Kind

13:14

we team technologists like data scientists

13:16

who want to volunteer their time alongside

13:18

social change organizations, be they government

13:21

agencies or nonprofits who have

13:23

a pro social mission, might be able to use data

13:25

and algorithms to do even more, and

13:27

we together they collaborate and kind of codesign

13:30

the solutions that they might foster a

13:33

better world. So some examples that we've seen

13:35

are exactly what you're talking about. There was

13:37

a project that a group did as a water district

13:39

in California, and the problem they faced

13:41

was when drought season comes, you know,

13:44

it's really hard to get water to folks. People don't have

13:46

water. That's obviously problematic.

13:48

You need drinking water and water to bathe,

13:51

etc. But more than that, the

13:53

cost of not getting them water is

13:55

really high because the only way

13:58

that they can get water to the places they don't have it is

14:00

to actually take a dump truck, drive

14:03

it up to some other reservoir, maybe over to

14:05

Nevada, literally fill it by hand and drive

14:07

it back. So you're also facing

14:09

like huge environmental costs, huge energy

14:12

costs. So they ask the question,

14:14

you know, could we figure out a way to predict

14:16

how much water demand there's going to be at

14:18

a more granular level so we can really understand

14:21

and ration more effectively. And

14:23

so we team them up with some data scientists

14:25

that come from everywhere from Netflix to

14:27

environmental science organizations, and

14:30

together they collected the data at

14:32

almost a block by block level, and they built

14:34

an algorithm that sort of takes that data in

14:36

and continually gives updates. Does water district

14:39

to say, hey, this is how much we think people

14:41

are going to use. Here's how much they've already used. Tomorrow,

14:43

you're probably going to see this, And they said, in the

14:45

first year of using this, they saved over twenty five

14:47

million dollars in addition to getting water

14:49

to people much more effectively. So

14:51

I think when you hear about cases like that those

14:53

are the kinds of examples that we want to kind

14:56

of platform and see more even the world where

14:59

within the confines of social

15:01

organization these data and algorithms that

15:03

can really drive real effectiveness.

15:06

Now your people are all doing this for good.

15:09

We've all heard about the kinds

15:11

of bias issues that have started

15:13

to turn up with predictive algorithms

15:16

of different kinds, and they seem

15:18

to get embedded just because

15:20

of the inherited unconscious

15:23

biases of the people who write

15:25

the algorithm. Absolutely, how do you avoid

15:28

recapitulating that problem again

15:30

with the projects you're working on? Such

15:33

an awesome question, and I think just

15:35

to comment on the challenge generally,

15:37

I think you really nailed it there. That

15:39

the challenge that we face is that

15:42

humans have been collecting data from

15:44

our activities that incorporate

15:47

unconscious bias, and so if you then have a machine

15:49

learn from it or you analyze it, you

15:51

write replicating that. So, while

15:54

I will not admit that we have a perfect solution, because

15:56

I mean we're sort of talking about the challenge of

15:58

bias and humanity, some

16:01

of the things that we really focus on is the

16:04

technology to us that we're building

16:06

is secondary to the outcome for

16:08

people. So, for example, it's not

16:10

exciting to us to build an

16:12

algorithm that helps a

16:15

homeless shelter triage people

16:17

to the right homeless shelters correctly just

16:19

because it's a cool algorithm. We only

16:21

care if at the end of the day, the ultimate success

16:23

metric that you know, a wide

16:26

range of inclusive folks are getting

16:28

housing is achieved. So

16:30

I want to say that first because I think one of the reasons

16:33

we see some of these biased challenges

16:35

rise up is that folks say, hey,

16:37

the algorithm is doing something. It's doing a thing I

16:39

want, like giving out sentences

16:42

in courts or you know, policing

16:44

folks, but without a question of and

16:46

how is it biased? Towards the end, you know, what's

16:49

it achieving. But the other thing we do is we work

16:51

extremely closely with our NGEO

16:53

partners who are on the ground and

16:56

who understand a lot of those challenges. And

16:58

so we'll actually do what we call a pre mortem

17:01

some other companies do, which is before we even start

17:03

a project, we'll say, okay, let's pretend we jump to

17:05

the end. Well, you know, basic

17:07

questions like how will this be maintained, who's

17:10

actually going to use this tool at the end of the day. But

17:12

then we'll also ask two questions, which is one,

17:15

what's the worst that happens if we fail? So

17:17

if you're relying on us to build, this

17:19

is not something we would necessarily build. But let's say someone

17:22

said, hey, we want a tool that predicts

17:24

whether you have cancer or not. Okay, well

17:26

that's pretty serious. And if we don't succeed,

17:28

are you stuck because you really

17:30

needed that and now your organization can't proceed.

17:32

That's important to know. But then we also ask

17:35

what's the worst that happens if we succeed? So

17:38

who is this going to affect? How would you know that it's

17:40

wrong? Right? Like, how would you know just because it's

17:42

chugging away making predictions? Is it doing

17:44

the right thing? Is it disenfranchising certain

17:46

groups? Could somebody use it to intentionally

17:49

target people who have cancer? We

17:51

ask a lot of those questions, and what's really

17:53

important us in that questioning is who

17:56

has the power and agency to both

17:58

understand the algorithm and change

18:00

the algorithm Because in the current landscape,

18:03

when tech companies build algorithms, it's not much

18:05

you can do. But you know, I don't have enough agency to know

18:07

how Facebook's news feed algorithm

18:09

works, nor can I really affect it much? But

18:12

that's not acceptable to me when you're bringing algorithms

18:15

into the public good space and this is actually

18:17

affecting folks lives. So those are some

18:19

of the questions we ask up front and really try to be rigorous

18:21

with our partners around oversight of and oftentimes

18:24

that's enough for us to not take on a project. It's

18:26

great that you're thinking steps ahead about

18:28

these projects, and your own

18:31

solvable is, ironically,

18:33

to put yourself out of business is to create

18:35

a world in which you don't need a data

18:38

kind to point people towards positive uses

18:40

of data. That's right, What would it take to make

18:42

that happen? And I guess

18:44

playing your chess game. What

18:47

happens when that happens. The day we close our

18:49

doors is the data. Every frontline social

18:51

change organization has the capabilities

18:53

to use data and AI the same way the big tech

18:55

companies do ethically and capably.

18:58

And so you know, our little slice of

19:00

that today is to bridge the gap

19:02

in getting the human capital, the talent, the

19:05

data scientists AI engineers to social

19:07

organizations. That sort of step

19:09

one is to show people the art of the possible

19:12

and really get some of those challenges solved. But

19:14

what do it take to do that? Long runs to think about what

19:16

are the problems and hurdles we're trying to overcome

19:18

with that model today, and they are that

19:21

in the social sector there isn't enough

19:23

awareness about what the technology could do or

19:25

where it would be applied. So we have to start with

19:27

that, and I think now increasingly you're seeing

19:29

more of more folks understanding

19:31

that, more companies talking

19:34

about doing data and AI for good. So

19:36

I feel like there's some progress there, But

19:38

if you go further, you have to think, well, how

19:41

would a government or nonprofit get

19:43

access to these resources in the long term,

19:45

And there I think there's going to be a

19:47

long term shift in getting funding

19:50

to move towards nonprofits

19:52

for things like data science and AI.

19:55

You're going to need maybe consultancies

19:57

that actually provide this service in

19:59

the social sector. There's lots of different

20:02

models for where that capacity could come from, but

20:04

I think the biggest things that we need right now

20:06

are that awareness of how could be used and then

20:08

the I say, the funding for ngox

20:11

to be able to hire a data sciences and

20:13

incorporate them into the work they do. Now.

20:16

When that happens, what happens. Oh,

20:18

I mean, I'd love to say that all

20:21

challenges that are stymied by

20:23

not having data science and AI are solved live

20:26

apply ever after. But actually, what I

20:28

think my most ambitious hope

20:31

for the world is that we could actually

20:33

tip the balance a little bit to where

20:36

the social sector is paving the

20:38

path for how machine learning

20:41

and AI could be used. I think we're so

20:43

built into this default model that business

20:45

and wealthy countries set the agenda

20:48

and everyone else kind of struggles to catch up and

20:50

imitate. We're talking

20:52

about a technology that is so fundamental

20:54

to humanity because it relies on data

20:56

about us. When we talk about AI,

20:59

it is like automating human processes that

21:01

I don't think that's something that should be just a business

21:03

application that is ported to the world.

21:06

There should be a place for us to say,

21:08

what does it look like when we apply the technology

21:11

to the better angels of our nature? What is

21:13

human based AI? What are the things

21:15

we care about? And I can't think of any other

21:18

place besides the social sector whose sole

21:20

mandate is to look out for humanity.

21:22

So my dream is when you bridge that

21:24

gap, when that's there. You could actually

21:27

have this voice from the social sector

21:29

itself saying what it looks like to have human based

21:31

ai Jick. Do you think about the

21:34

training of data scientists.

21:36

I sometimes think we're just missing

21:38

the intersection between moral

21:41

philosophy and computer

21:43

science. You know, the people who are majoring

21:45

in college and electronic engineering

21:48

aren't reading much Kant, and the people who

21:50

are reading Kant don't understand much

21:52

about computer programming, you know, And

21:55

in a way, the problem is that the people at

21:57

these tech companies don't have

21:59

a different kind of background in literature

22:02

and philosophy and history to

22:04

think through the implications of what

22:06

they're building the way you clearly are

22:09

thinking through those implications. I think

22:11

it's a really great point that when

22:14

wielding the technology, it's really

22:16

important to have a

22:18

very varied sense of skills

22:21

somewhere in the conversation. And increasingly

22:23

you're seeing data science and tech curricula

22:25

incorporate ethics training into

22:27

their courses, which I think is great. In the same

22:30

way that I'm not a historian myself,

22:33

I feel like physics went through this reckoning with

22:36

the ethics of what was being built when they went from

22:38

the joy of all energy and nuclear

22:40

power to the realizations of the downsides of

22:42

the nuclear bomb nuclear weapons. So

22:45

I think you're going to see that similar shift, which is which

22:47

is great, But you know, I think what

22:50

your question raises actually a bigger point to

22:52

me, which is who holds

22:54

the responsibility for the ethical

22:56

applications of this technology? And

22:59

I'll just say, while I would love to

23:01

see, you know, ethical code around

23:03

data science, it's a lot

23:05

of responsibility to say that engineer

23:08

x it has come out of college engineering

23:11

college for two years and is working at big tech

23:13

company and gets asked by

23:15

their boss to build something fairly

23:17

benign, like I upgrade

23:20

to their their GPS system that

23:22

recommends routes you can walk that

23:24

avoid crime ridden areas. I say,

23:26

here's an algorith build that. Well,

23:29

number one, that's not necessarily a bad thing to

23:31

builds not like you know, it's not as black and white

23:34

as some people may feel about building a weapon or

23:36

something. But of course, if you sort

23:38

of play the game through, if

23:40

everyone were using an app that avoided crime ridden

23:42

areas, probably end up with some sort of digital

23:45

segregation. So number one, there's already

23:47

long range effects that you'd have to anticipate.

23:49

But more than that, It also relies on that,

23:51

you know, second year engineer to say, hey

23:54

boss, yeah, I'm not doing that. You know

23:56

this is I'm quitting, which, given

23:59

you know people's career paths and

24:01

the money associate with these jobs, is a

24:03

big ask. So I would say

24:06

it's not just about the technologies.

24:08

I think the question is, you know, how do we

24:10

share that responsibility? Is it the technologist

24:12

to make this call? Was it the manager said we want

24:14

to build this feature? Was it the constituents

24:17

would be affected by that? Is a government

24:19

to come regulate. I don't think there's any one answer,

24:22

but I do think the frame that people have

24:24

I'm hearing more in the public

24:26

right now around technologists need to know

24:28

the ethics, I think is missing the bigger

24:30

picture that that alone isn't the right responsibility

24:33

model. In my mind. You have two very

24:35

different ideas of capitalism, right. I

24:37

mean, there's an older idea

24:40

that government sets the rules,

24:42

tells you what you can and can't do, and that businesses

24:44

should obey the law and regulation

24:47

but go be very free to do what they want.

24:49

Within that, the newer

24:52

model suggests that the

24:54

businesses themselves have a higher degree

24:56

of social responsibility, and

24:58

it's not enough to follow the rules

25:01

that they have to be thinking about outcomes.

25:03

Look, I would love to live in a world where

25:06

business and social

25:08

outcome were somehow linked,

25:11

where the fact that businesses were accountable

25:14

somehow to at least not doing harm,

25:16

if not improving human life. That would be a really

25:18

great intersection. Call

25:20

me a cynic, but we're not really currently

25:22

set up for that. The incentives aren't there.

25:25

In my mind, businesses are still

25:27

held mostly to the bottom line, even though we are

25:29

seeing some increased interest

25:32

in social entrepreneurship, where businesses

25:34

may have a double bottom line, one that's

25:36

monetary and one that's social, or

25:38

new structures like b corps that actually

25:41

say, hey, we are committed to some social cause. But

25:43

I think it's a lot to ask of

25:46

a company. And as much as it's a nice

25:48

idea of a future of capitalism, it's certainly

25:50

not the rule or the law. And

25:53

so I don't think that's

25:55

going to be the sole model that brings us to a world

25:57

of pro social technology and AI.

25:59

If for no other reason then certain human

26:02

needs are inherently

26:05

cost ineffective, I would say to solve

26:07

at least currently if people could cry those if every

26:09

social problem were able to align perfectly

26:11

with a business needs, be in great shape.

26:13

But when it comes to housing

26:16

the homeless or making sure that people have

26:19

food to eat, that is a

26:21

difficult challenge that I don't see an immediate

26:23

market solution too, and so I don't

26:25

think even the best intention companies could survive

26:28

in a market based world trying to solve that

26:30

problem. I mean, Google, which is still

26:32

the first and best known data

26:35

company essentially has held

26:37

out this promise that we're

26:39

going to make a lot of money using data

26:42

commercially targeting advertising,

26:44

but we're going to use a lot of what

26:46

we make, or at least some of it

26:49

in a kind of philanthropy. We're going to try

26:51

to create some of the kinds of solutions

26:54

you're talking about that aren't driven

26:56

by the profit motive. Does that work look

26:59

like? I said, One of the big challenges we

27:01

face, I think in the social sector right

27:03

now is the lack of funding for innovation

27:06

for your technology. And so

27:08

if company are going to offer

27:10

that great netwin,

27:13

do I believe that the world's biggest

27:15

challenges will be solved

27:17

on the you know, philanthropic efforts

27:20

of large companies that

27:22

I'm not so hopeful. I think there.

27:25

I still wonder where are the folks for whom

27:27

the mandate is solely pro social,

27:29

you know, for governments or again nonprofits

27:32

or civic organizations whose very

27:34

guiding mission is to make sure that human prosperity

27:37

is enhanced. There's

27:39

a little bit more of a direct line there. And so that's

27:41

why I think it has to be a combination

27:43

of the two, and why we focus so much on

27:45

saying instead of trying to bend

27:48

the Googles of the world to you

27:50

know, being in charge of clean water, which frankly I

27:52

think is really not not the way you want to go. Where

27:55

the you know, the clean water organizations of the world

27:57

who just need that same technology to be ten hundred

27:59

times more effective. What are some things

28:02

listeners to this podcast might be able

28:04

to do to work towards the kinds

28:06

of solutions you're thinking about. Well,

28:08

the great thing about this cross cutting

28:10

technology is that everyone has a role to play

28:13

in creating this future vision of more

28:15

social and positive AI. Well,

28:18

first, I would say, if you're a technologist who works with

28:20

data and you want to give your time

28:22

and energy back, come aboard. There's

28:24

a whole movement of folks doing this work. Whether

28:26

you want to come work with us at Data Kind and work on projects

28:29

pro bono, or with many of the other organizations

28:32

like Driven Data, Data Science

28:34

for Social Good, CODE for America

28:36

who take technologists and apply them to social

28:38

problems, come aboard. There's no reason to

28:40

wait. And increasing Link asked the

28:42

company you work for if there's

28:44

opportunities to give back, because we see more tech companies

28:47

do that. But if you're not a data scientist,

28:49

non data scientist, I would

28:52

say, yeah, I have to first give a shout out to

28:54

anyone of the funder or donor world. One

28:56

of the big gaps here is that there

28:58

isn't enough funding for technology and innovation

29:00

in the social sector. So I've been very

29:02

impressed with the efforts of Rockefeller Foundation

29:05

and MasterCard Impact Fund and

29:07

others who are giving big amounts

29:09

of funding to data and AI and social

29:11

good to bring it on. We need more of that for

29:14

this happen. But very lastly,

29:16

if not a data scientist and you're not

29:18

a funder, I would say there's a

29:20

huge opportunity to get involved in

29:22

just understanding what this new technology

29:24

can do. Ciicero had a quote that

29:27

you should take an interest in politics, because politics

29:29

is definitely going to take an interest in you. And

29:32

I feel exactly the same about data and algorithms.

29:34

They're going to take an interest in all of us. In fact,

29:37

they're shaping our lives already today. Maybe

29:39

the reason you're listening to this podcast is because an algorithm

29:42

recommended it to you based on your previous listening

29:44

habits. And so if these tools are

29:46

going to be shaping and visibly

29:48

shaping our decisions, then

29:51

it's all the more incumbent on us as

29:53

society to understand

29:56

what the ramifications are, where

29:58

it's showing up in society, and

30:00

how we might have some agency over the role

30:02

we want it to play. I think so much

30:04

of the reason you hear so much negativity today

30:07

is because we don't understand it well

30:09

enough and we don't have any agency to change it. So

30:11

our only options are to shrug and say, well,

30:13

I guess that's going to be the way it is, or

30:15

to rail against it and say this is bad. But

30:17

if we could get to a place where we had

30:20

call it algorithmic literacy. Not

30:22

everyone needs to code, but if you just understand a

30:24

little more about it, then I think we'd progress

30:26

towards a society where we felt like

30:29

we had a more control agency over

30:31

how we work with the machines instead of against

30:33

them. That's a great point. And I have

30:35

to ask you for a reading recommendation. If

30:38

people need to get educated, what should they read.

30:40

What's a thing or two they should read to

30:42

get more sophisticated about data.

30:45

So the best thing I think you can read

30:47

are some of the blogs that actually talk

30:49

about the state of the space today, because

30:52

it's changing so much that you know there's no

30:54

one book that's going to capture it. Yeah. So some of the

30:56

ones I love are the company

30:58

O'Reilly O'Reilly dot com.

31:01

They have a feature on data

31:03

and AI that's a weekly newsletter that comes

31:05

out talking about everything from the interesting

31:07

innovations and AI to what

31:10

kind of privacy concerns are

31:12

in the space today, and it's very readable for

31:14

a common audience. I think that's one of the most interesting

31:16

ones. I would also read Data

31:18

and Society's newsletter. They are a group

31:20

here in New York who are really tackling

31:22

the question of what does it mean to have data

31:25

and algorithms in society. They have some

31:27

really great accessible writing there The

31:29

other thing I would say is if you have

31:31

the privilege of living near a

31:34

medium, miss or big city that has a

31:36

meetup community. There are

31:38

tons of data science AI meetups

31:40

where people go and just talk about what's going on

31:42

in the space. And I always recommend

31:44

that people drop by at least one because

31:46

if you see it and feel it and here people

31:49

are talking about you don't have to understand,

31:51

you know, if there's any math on the board, but just you

31:53

almost immediately, it creates

31:56

a states where people walk and go, oh,

31:58

I actually see what this is all about. So

32:00

I would say if you happen to be a checkout meetup,

32:03

dot com or any of those communities. The

32:05

data scientists AI folks are very friendly

32:07

and I know you'll have a great time, if not an

32:09

educational one. Terrific. Well, Jake

32:12

Probi, thanks for joining us Unsolvable My

32:14

pleasure. Thanks so much for having me reasons

32:18

for hope all of this potential being

32:20

harnessed to improve people's lives,

32:22

the really big stuff. Although

32:25

my ears certainly did prick up when Jake mentioned

32:27

Twitter for pets, as did my dog's

32:30

ears. She has been dying to get online

32:32

and really drag other dogs anonymously,

32:35

of course, but both myself

32:38

and my dog are pleased to see what

32:40

data Kind has actually managed to do so

32:42

far, creating algorithms that

32:44

have helped transport clean water more effectively,

32:48

informed government policy that protects

32:50

communities from corruption, and detected

32:52

crop disease using satellite imagery.

32:55

Jake and his team and all those volunteers

32:58

are leveling the playing fields and you can

33:00

help too. Read more about data

33:02

Kind and how to get involved at

33:04

Rockefella Foundation dot org. Slash

33:07

solvable. Solvable

33:10

is a collaboration between Pushkin Industries

33:12

and the Rockefella Foundation, with production

33:14

by Chalk and Blade. Pushkin's

33:17

executive producer is Mia LaBelle.

33:19

Engineering by Jason Gambrell and

33:21

the fine folks at GSI Studios.

33:24

Original music composed by Pascal

33:26

Wise. Special thanks to Maggie

33:28

Taylor, Heather Faine, Julia Barton,

33:31

Carlie Migliori, Sheriff Vincent,

33:33

Jacob Weisberg, and Malcolm Gladwell.

33:36

You can learn more about solving today's biggest

33:38

problems at Rockefella Foundation

33:41

dot org. Slash Solvable.

33:43

I'm Mave Higgins, Now go solve

33:45

Itt