Robert Bjarnason about the Citizens Foundation and how technology supports participatory democracy
Ep. 16

Robert Bjarnason about the Citizens Foundation and how technology supports participatory democracy

•

Episode description

Robert Bjarnason is an early Internet entrepreneur and is co-founder of the Citizens Foundation in Iceland 2008. This was at a time when the financial crisis hit the world and trust in the parliament of Iceland was at a historical low. The idea was to start a civic tech non-profit to help create technology that enables better decision-making in society.

Download transcript (.srt)
0:00

Welcome to another episode of Democracy Innovator

0:02

Podcast. Our guest today is

0:05

Robert Bjarnason,

0:08

the co-founder of the Citizen Foundation and

0:12

many other things,

0:15

but I think you will tell

0:18

us something about this in this episode.

0:21

Thank you for being here. Thanks for inviting

0:23

me.

0:25

As a first question,

0:27

what is the Citizen Foundation and

0:30

how did you start it?

0:33

The Citizen Foundation is a nonprofit.

0:35

We're actually two nonprofits.

0:37

One is registered in Iceland and

0:40

also one is American in the US.

0:42

It started in 2008. I was

0:45

actually living in London

0:46

at the time when

0:48

the financial crisis hit Iceland.

0:51

Iceland was a pretty vulnerable victim, and

0:54

over a period of time, all

0:58

the news was about Iceland

1:00

being bankrupt and so on.

1:02

You know, going by taxi

1:03

in London,

1:05

talking to the taxi drivers,

1:08

"Are you from Iceland? So unlucky,

1:09

your country is bankrupt."

1:11

Well, actually it didn't turn

1:13

out really that bad in the

1:15

end, but at the

1:16

time it was an overwhelming feeling

1:19

in Iceland where

1:22

trust in government and parliament

1:25

plummeted down to six percent.

1:27

This is one of the oldest parliaments

1:31

in the world, founded in the year

1:32

930, so like over

1:34

a thousand years old. Over a

1:36

thousand years of pretty good trust

1:38

in the parliament, but it plummeted

1:39

from sixty-seventy percent to

1:41

six percent in two weeks.

1:43

Then it was just overwhelming -

1:45

this idea that we weren't really

1:48

in control over what's going on anymore.

1:51

It was certainly in Iceland,

1:53

but also in wider society

1:54

in the rest of the world,

1:56

at least the Western world

1:59

like in the UK where I was.

2:00

So we had this idea -

2:01

this group of people,

2:04

mainly my partner and I

2:05

who started the first internet service company

2:07

in Iceland in 1993,

2:10

plus some other people from our network.

2:12

We had this idea to start

2:15

a civic tech nonprofit to

2:17

help create technology to enable better decision

2:20

making in society between governments and

2:22

citizens. We've always been using a lot

2:24

of open source software

2:26

and promoting and creating open source software.

2:28

The basic idea was:

2:32

how can we enable, by using

2:35

technology, the internet, and AI and other things -

2:39

how can we help governments

2:42

and citizens make better decisions in an

2:43

increasingly complicated world?

2:46

I was wondering, when you had

2:49

the idea that technology could help

2:52

people make better decisions -

2:56

you mentioned 2008, but you also said

2:58

that you opened an internet provider.

3:00

Was that also connected to

3:03

the idea that the internet could bring

3:08

more discussions?

3:10

Yeah, absolutely. I was twenty-one when

3:12

I started the internet company in '93.

3:16

The internet existed since 1987 really -

3:20

the text-based internet - but then the

3:22

web came and we started this

3:23

internet company.

3:27

There was a lot of idealism about how

3:29

this increased communication could give people

3:32

a stronger voice in society

3:34

and even in a utopian way,

3:36

how the internet and cyberspace could really

3:39

transform social power structures

3:41

and make society fairer and more equal.

3:44

So yes, I was definitely a

3:45

part of that. That's also why we

3:47

started that internet company.

3:49

I'd been helping schools connect to the

3:52

internet. I was at the Teachers

3:54

University, and I

3:57

downloaded the first web browser called Mosaic.

3:58

I'd been on the text-based internet

4:00

for years,

4:01

and I tested it and was like,

4:04

"Wow, this is the future."

4:07

We actually downloaded the first

4:08

public Linux operating system called Slackware

4:12

on floppy disks. That was all

4:15

for free - Linux, you know.

4:18

We set up telephone lines

4:20

in my friend's living room.

4:21

So I'd always been interested in

4:23

using all this open source software as well.

4:24

So the idea in 2008 -

4:26

opposed to this idea of

4:29

the internet - I mean, the internet was not really

4:31

delivering. Yes, it's given a lot of people

4:33

voice, but when it comes to actually giving people

4:36

a voice that can influence decision making,

4:39

there was almost none.

4:40

But also, just in

4:41

general, how open source

4:43

community-based things can really work.

4:45

Linux showed that, the internet

4:46

showed that you can actually have grassroots open source

4:49

solutions that really make big changes

4:52

in society. That's also the reason why we

4:55

decided that since we're going to be

4:58

working with citizen engagement to build

5:00

up trust, we couldn't be a

5:02

for-profit company. We had to be

5:04

a nonprofit organization because

5:06

otherwise we wouldn't have the right

5:09

alignments.

5:10

In your opinion, is the internet

5:13

still a space of freedom, or

5:14

can it also be a space of surveillance

5:17

in some way?

5:20

I think the internet is still a

5:22

place of freedom in a way,

5:23

but it's also basically a place of surveillance.

5:25

There are different types of

5:26

surveillance in different places of the world.

5:28

In some places it's highly

5:30

monitored and not really free at

5:32

all for free thought.

5:33

But I think the internet, like everything else,

5:35

just reflects the society

5:38

that it's embedded in.

5:41

So what are the

5:42

main solutions that you propose to

5:44

this distrust that is missing regarding

5:46

classic democratic systems -

5:48

Western democratic systems?

5:49

When we started, we felt like,

5:51

"Okay, so how can we use the internet?"

5:53

We started thinking very quickly about

5:55

e-democracy, collective intelligence.

5:56

There were quite a few people -

5:58

academics and others - trying different things.

6:00

The basic idea is

6:02

that if you have a

6:02

complicated decision, and in the modern world

6:04

we have increasingly complicated decisions

6:07

with competing demands -

6:11

what is considered a good decision today

6:14

maybe is not exactly the same as

6:17

twenty years ago. The world is

6:19

evolving, it's not static.

6:22

The idea is if you're doing

6:24

a complicated decision or decision-making process

6:26

that's going to affect a lot of people,

6:28

then actually by reaching out

6:29

at the right time when you're creating the

6:32

policy to citizens to actually get

6:33

information - this basic principle is

6:36

that we're going to make better decisions

6:37

with better information.

6:38

I think few people

6:40

can argue against that - it's

6:42

even mathematically proven that

6:43

more information is going to give you

6:45

better decisions.

6:47

That's the core of the

6:49

idea - the utilitarian part.

6:53

But obviously there's also the democratic

6:57

angle of it. What makes a good

7:01

democratic decision is something that is going to

7:04

be balancing the competing interests

7:08

of the voters and the people who

7:10

are living in a society that

7:14

the rule or regulation or law

7:16

or decision affects.

7:17

This can be small things

7:19

like participatory budgeting - are we going to put

7:22

a playground there or an

7:23

outdoor hiking park? There are lots of

7:26

good examples of participatory budgeting

7:28

all around Europe and the

7:31

world. But also policy

7:33

like educational policy, traffic policies -

7:37

all sorts of policies where you have many

7:39

different stakeholders.

7:43

For example, the Scottish Parliament, where

7:43

the platform and this idea is embedded into

7:45

the committee system of the Scottish Parliament.

7:48

They've been doing really groundbreaking

7:50

work, leading the world in many ways

7:53

when it comes to engagement.

7:54

They have an engagement

7:55

unit and they take it very seriously,

7:57

very professionally. They offer

7:59

a service to the different committees

8:01

to say, "Oh yeah, we have this issue,

8:03

do you need information from the people,

8:05

from the public, before you take this decision?"

8:08

And then the committees say, "Oh yes, we need

8:10

information on X, Y, and Z,"

8:12

and then that gets

8:12

implemented. It's like a way of

8:15

actually delivering real

8:17

value to the elected

8:21

representatives in the committees by getting

8:23

good information to them.

8:25

This is used quite many

8:26

times. It's just examples of how

8:28

this kind of engagement can

8:30

both lead to better decisions

8:32

but also

8:33

help empower people democratically

8:34

because they can influence the decisions

8:36

themselves by taking part.

8:38

I was thinking - so now

8:39

we have representative democracy.

8:41

Do you think that in the future

8:43

with this kind of deliberative

8:45

tools like participatory budgeting

8:47

and all these new tools that maybe use AI,

8:49

we could have some different kind

8:51

of political system? Maybe without

8:53

representatives, maybe still with some

8:55

representatives or professional politicians?

8:59

Obviously anything is possible

9:00

in the future, but I think

9:02

democracy, law, and everything -

9:05

how it all connects together,

9:07

the norms that people are used

9:09

to - there are all sorts of different

9:11

things that need to change

9:13

together for some really major changes

9:16

to the system.

9:19

One of the things that we've been very much

9:21

against in our work for

9:23

seventeen years since we started this is

9:26

direct democracy. We have

9:28

taken clear stances that there are

9:30

lots of decisions where direct democracy

9:32

can't work today.

9:33

I think that's one of the things -

9:36

this idea of proxy voting, for example,

9:40

where you delegate your vote to certain

9:41

parliamentarians. There are issues with direct democracy.

9:44

We have referendums obviously,

9:45

and Switzerland is obviously famous for that,

9:48

but we have taken this stance

9:50

against direct democracy for several reasons.

9:53

One is security -

9:53

how secure are you doing it online?

9:56

We're lucky here in Iceland,

9:58

and also Estonia and some other Baltic countries,

10:00

where a lot of people have

10:02

electronic IDs that enable any

10:03

type of direct democracy.

10:05

But if you don't have reliable

10:07

electronic IDs, you shouldn't be doing

10:09

direct democracy where people vote on insecure platforms

10:12

about something important, because

10:14

that's going to totally undermine

10:15

democracy. It's going to create distrust

10:17

in democracy if people are voting

10:20

on platforms that are not secure.

10:23

That already limits lots of

10:24

places from really doing any

10:26

reliable direct democracy.

10:28

But we also draw the line

10:31

at participatory budgeting things

10:34

where people are voting for priorities.

10:37

In Better Reykjavik - our main project

10:39

that's been running in the city of Reykjavik

10:41

for fifteen years now -

10:43

people have a chance to put forward

10:45

ideas about what to do

10:48

in their neighborhoods. Seventeen hundred ideas

10:49

came in a couple of years ago.

10:52

Six months later, the city

10:55

looked at them, professionals costed them,

10:56

and then people can vote with

10:58

their electronic IDs on what ideas

11:01

are going to be implemented in the

11:03

neighborhoods. But if it was about

11:06

policy, then we have a

11:08

totally different dimension where

11:10

the power actually lies.

11:15

The thing is that the

11:17

power in online things is

11:19

controlled by who knows about it

11:22

and how you control who knows about it.

11:25

You do that through social media,

11:28

and how do you control who sees what

11:30

on social media? You pay for it.

11:35

So basically

11:38

anything you do online

11:40

for direct democracy - we're voting on this

11:43

issue or that way - for us

11:45

is not in line with our values.

11:47

We always recommend against it because

11:50

even if you have electronic IDs,

11:51

if it was a controversial issue

11:53

and it's all online,

11:55

it's just too easy to

11:56

manipulate. That's just the problem

11:57

of referendums in general - referendums

11:59

can go all sorts of different ways

12:02

recently, as we've seen.

12:04

But having said that,

12:06

I think what we can see

12:07

in the future is that

12:10

AI is going to help empower

12:12

the organizational side of

12:15

governments because often the reason

12:18

why there isn't participation is because

12:20

of lack of resources and

12:22

lack of planning capabilities to actually incorporate

12:24

the feedback. I think AI

12:28

is really going to empower that

12:29

part - it's going to increase the capacity

12:30

of government to actually

12:33

work through good decision making

12:35

exponentially, not just in the

12:37

next two years.

12:39

But then also with AI,

12:41

it's going to help empower

12:42

citizens. For example, with

12:45

doing a new iteration of Better Reykjavik

12:48

later this year - we have over forty thousand

12:49

registered users - we're going to have

12:51

an AI system that is going

12:54

to allow people to have

12:55

an AI agent that's basically going to be

12:58

watching out for their interests in the

13:01

city. Depending on the

13:03

subjects they're interested in,

13:04

it's going to be like your agent

13:06

watching out for your interests.

13:09

I think both of

13:11

those aspects - both in terms

13:13

of more capacity for governments

13:15

and also more help for citizens -

13:17

that's going to really help with the quality

13:19

and quantity of citizens

13:21

working with government and AI for better

13:24

decisions.

13:26

I was thinking about power

13:29

and information. If I don't have

13:30

information, I also don't have the

13:31

power to choose to do A or

13:34

B. Then there is this question

13:36

about the lack of participation -

13:38

if people don't have the information,

13:40

they can't actually participate.

13:42

If they had

13:44

information about a certain person,

13:46

maybe that person could see about

13:47

changing something in the neighborhood where

13:51

they live. I think this is a

13:54

cultural problem related to

13:56

education, because maybe in school

13:57

we are not taught

13:59

about how we should participate

14:01

and how it is important.

14:03

Absolutely. One of the things

14:04

that we've done over the years is

14:07

that especially after mobile phones

14:09

became popular like in 2013,

14:11

2014, 2015, when everybody

14:14

had smartphones, was to lower

14:16

the barrier to participation

14:19

both through technology and

14:22

design. For example,

14:24

when you are

14:24

visiting a civic project on the Scottish Parliament

14:25

or something, the first thing you see

14:28

is a nice image, then

14:30

there's a very short description of what

14:33

to expect, and then

14:34

you can see more through a

14:38

"read more" button. It has a very

14:41

simple user interface which we also have

14:44

designed for deliberation.

14:44

Instead of when people see

14:47

an idea asking them to

14:49

comment on it, we ask them to

14:51

help us out - help us

14:53

find the pros and cons.

14:55

We ask them to help us find the best

14:57

arguments. I think that's an

14:59

issue - people need enough

15:01

information, but also how

15:04

the information is presented to them

15:04

and at what level you break down

15:06

information. For example, we did this

15:08

with the government thinking about constitutional

15:11

changes in Iceland. We worked

15:13

with them on a Better Reykjavik platform

15:15

where they basically had to

15:16

break down the constitution into

15:18

parts - like four different things they wanted to

15:20

change. You could have put a

15:23

big PDF with all the legal text

15:24

and asked people to give comments,

15:27

but we knew that would never work.

15:29

So what we did on our platform was

15:30

create pages where you have

15:32

the three top-level topics in

15:35

very short text. You click on

15:36

each one and then there are four things

15:39

that the government wanted to do in the constitution

15:42

connected to each topic - very clear,

15:44

almost lowering the barriers

15:45

so almost everybody could understand it.

15:47

That simple change, compared to

15:49

the PDF with the actual law proposal -

15:53

that's like a barrier that

15:55

ninety percent of people can't deal with.

15:57

We lowered it so that ninety percent

15:59

of people can process it because

16:01

it's been broken down like that.

16:02

That's really important. But I think

16:04

when it comes to the

16:05

information part itself, I think

16:06

AI is already helping people

16:08

who are using ChatGPT or

16:11

Claude or whatever. It's both

16:13

doing searching for you, it knows your preferences,

16:15

it's like on your side in a way.

16:19

If you get a

16:20

subscription to ChatGPT, at least now

16:23

it has no advertisers, it's just you

16:25

and the model, so in theory

16:27

it's going to be on your side,

16:28

knowing your preferences.

16:30

What we're doing with our

16:32

AI agent for Better Reykjavik

16:36

for your priorities is that

16:40

it's not just looking out for your interests

16:42

in the city, but we are actually going to

16:44

build it in an open source way

16:45

where you can use open source models,

16:48

building it in a way where

16:51

you can basically tell

16:52

the model, "Those are my

16:54

interests and the things in the city,

16:56

my needs - I'm a parent,

16:59

I have those children with those abilities,

17:02

I have a car," whatever.

17:05

Then the AI will help you

17:05

find the information.

17:07

I think about not having the information -

17:09

if you go back thirty years ago,

17:12

you had some people who

17:14

would read all the newspapers every day

17:17

and have all the information,

17:18

and some people who didn't read

17:22

papers at all and

17:23

wouldn't be up to date on anything.

17:26

But today it's totally different -

17:30

information is coming from everywhere,

17:32

some of it is true, some not.

17:35

So having the information has changed

17:37

a lot because there's just so much information.

17:39

I think AI is the solution to help

17:42

us deal with that by

17:45

understanding your priorities.

17:48

Is Your Priorities the

17:51

main platform that you're working on?

17:54

But before asking you about it,

17:57

would you like to share something

18:00

about your professional background -

18:02

academic, but whatever?

18:04

Very short on my

18:06

academic record, but I

18:08

started - my father's an electrical engineer

18:11

and my mother's a doctor.

18:15

My father was early into computers,

18:18

and I guess I was

18:24

hyperactive, so he got me programming

18:29

at nine. I had a computer

18:32

called the Oric-1 microprocessor -

18:34

it was like a training computer for kids.

18:36

I got interested and really

18:39

spent all my energy on it.

18:40

I sold my first piece of software

18:42

in 1984 - a teletext

18:43

information booking system that would

18:45

scroll information. It was

18:46

used by the national government

18:48

for meteor strike information in

18:51

1984. I sold my

18:53

first game program in 1987

18:56

in a programming language called Prolog -

18:58

an expert system. Then I

19:01

took a break from computers.

19:04

I was involved in seven

19:06

feature films,

19:07

spent a year as a substitute lighting

19:08

director for the opera,

19:10

club promotion, putting on raves,

19:12

things like that. But then

19:14

I got back into it.

19:15

I was at the education

19:17

network, which was a pioneering

19:19

internet network for teachers,

19:21

helping connect schools to the

19:23

internet. That was when I saw

19:25

at the Teachers University the

19:27

Mosaic browser, and we

19:29

started this ISP.

19:31

In '95, I started the

19:33

first of two ISPs

19:34

in Denmark. Then I was

19:37

mostly traveling between there

19:39

and San Francisco where I

19:42

was working at a video game company.

19:44

I met some great people,

19:45

including my brother.

19:47

Then I got into really exciting

19:48

technology back then - chat bots.

19:50

In 2001, I met Lynne Hershman,

19:53

who had just done a movie

19:55

called Teknolust. She came

19:57

to me and said, "Hey, can you put

19:59

her on a Palm Pilot?" So we

20:01

did that. It was actually an exhibition

20:02

in San Francisco Museum of Modern Art

20:06

for a year. People

20:08

could beam the chat bot

20:09

into their Palm Pilot using infrared light.

20:11

Then I moved to London

20:14

after the dot-com crash,

20:15

worked on video games,

20:17

mobile phone games, and many other things.

20:20

At thirty-six, I decided to

20:21

take a break from the video

20:23

game industry and start a fintech

20:24

company working with hedge funds

20:26

and trading platforms.

20:28

At that point in 2008,

20:31

when this financial crisis happened,

20:33

frankly if I had still been

20:35

in the video game industry,

20:36

I probably wouldn't have started the Citizen Foundation.

20:39

But being in the financial world

20:41

gave me the opportunity to

20:42

make money so I could start the foundation

20:45

myself. Also, this idea -

20:49

if I can use technology

20:50

and AI to help rich people

20:53

make more money, why can't I

20:54

also use technology to make

20:56

society better and use technology

20:59

for good? Since then it's been

21:01

the Citizen Foundation for seventeen years.

21:03

We developed Your Priorities

21:04

very early on. We also

21:06

started a collective intelligence

21:09

AI platform three years ago

21:10

and started to work with

21:12

Northeastern University's Center for Social Impact.

21:14

We've been using AI and

21:18

large language models since 2019,

21:20

so I've been living in the

21:22

AI world for a long time.

21:24

The capabilities of what AI

21:26

now has have been growing really fast

21:28

the past five years,

21:29

so we've been slowly

21:30

transforming all our platforms

21:31

to be AI-first.

21:34

I read that you also

21:37

started a pirate radio?

21:39

Yeah, that's because my

21:41

father's a ham radio operator.

21:42

When I was seven or eight,

21:43

I convinced my father to

21:45

let me use his FM

21:47

transmitter with my cassette player

21:48

and a little microphone.

21:49

I had my cassettes - I

21:50

thought I was doing like a favor

21:53

with a five hundred meter or

21:54

one kilometer radius where all

21:56

my friends were listening in their cars.

21:59

I was playing awesome music

22:00

of the time. This was back

22:02

in 1979 or something,

22:04

a long time ago.

22:07

I was thinking - 2008 was

22:09

a significant year because of

22:11

the crisis, but Bitcoin was

22:13

born in 2008. There is

22:16

a lot of speculation behind Bitcoin,

22:17

but also some ideology.

22:18

What are your thoughts about

22:21

Bitcoin or blockchain?

22:22

To be honest, I've

22:24

never really got into it.

22:26

I've stayed away from that.

22:29

Every time people have asked me

22:30

to support liquid democracy

22:32

or whatever using blockchain,

22:35

I've stayed away from that.

22:37

I like the ideology - this

22:39

early internet, cyberspace stuff

22:42

where we can actually create

22:45

other power structures that are

22:47

more fair. I like that,

22:49

but it hasn't really happened

22:51

like that. It hasn't really

22:54

been of any use for

22:56

at least what I'm doing.

22:59

I think it's a good idea,

23:02

but I just haven't seen

23:04

any applications that have convinced me.

23:06

Because even though there's this

23:08

decentralization thing - once again,

23:12

it's proven that yes,

23:15

in theory you can have it

23:19

be untraceable and decentralized,

23:21

but then humans are humans

23:24

and they're going to use exchanges

23:26

all over the place. There are

23:29

ways of looking at and

23:31

knowing what's going on.

23:33

Even criminals, I guess,

23:35

but it's not really fulfilling

23:37

the original promise - or not

23:40

the promise, but the original things

23:43

people projected on it.

23:46

Humans are humans - I think

23:48

that's a key phrase.

23:50

The humans that are

23:52

using Your Priorities - are they using it

23:54

for which kind of

23:56

uses? Can you share maybe some

23:59

use cases?

24:01

So basically, most of the

24:03

use is when there

24:06

is a project going on.

24:08

For example, right now there's

24:11

a consultation with the Scottish

24:14

Parliament about ADHD - Attention

24:16

Deficit Hyperactivity Disorder.

24:17

They started like a couple

24:19

of weeks ago, already got

24:21

like seven hundred responses.

24:25

It's anonymous. Thousands of people

24:27

have come to learn about it -

24:28

they advertise it somewhere else

24:31

and people come and participate.

24:33

In a city context,

24:35

there's collecting feedback

24:36

on a new traffic policy,

24:37

collecting ideas. We have

24:40

this feature called deliberation -

24:41

a debate solution where we

24:43

really manage to encourage people

24:45

to find the best arguments.

24:47

Back in 2009 when we

24:50

did our first pilot with

24:52

the Shadow Parliament in Iceland,

24:55

we used a regular

24:57

commenting system like on

24:59

YouTube or Facebook - just

25:01

like a thread of comments.

25:02

In the first evening of

25:04

the platform about some policy,

25:07

we had these people in

25:08

a horrible personal argument

25:10

on the first evening.

25:12

We lost the platform users -

25:13

they were going at each

25:16

other personally.

25:16

I was looking at it like,

25:18

"Oh, this is not going

25:21

to work." But what we

25:24

came up with - we got

25:26

inspiration from parliamentary debates

25:29

and other university debate

25:32

structures. Also my experience

25:34

in the video game industry

25:36

is that

25:36

when you are building

25:38

a user experience like a

25:41

website or an app,

25:43

one secret is that you

25:45

are really in control of

25:47

the experience. When you have

25:50

a computer game, the player

25:52

thinks they're in control,

25:54

and they are to an extent,

25:55

but it's mainly the game

25:57

developer who controls exactly

25:59

what the experience is going to be.

26:01

The same applies to any

26:04

digital platform.

26:05

So I had this idea -

26:07

lots of people play video games

26:08

and think about how to

26:10

run the game. Could it

26:13

be an idea to mix

26:15

video games with participation

26:17

and debates, where people are

26:18

actually debating about the problems

26:21

of our society?

26:22

Absolutely. I think

26:23

gamification - when you take

26:25

small game elements -

26:26

but we also have this idea

26:28

for an actual game.

26:30

Instead of just asking for comments,

26:31

by having people help us

26:33

find the best arguments

26:35

in a project,

26:39

we get people putting both

26:40

pros and cons,

26:42

where the same person writes

26:44

both. There's almost never

26:46

any arguments just by

26:49

changing into this new

26:52

structure of design.

26:54

Also, you can't comment directly

26:57

on somebody's point - you have to

27:00

do a counterpoint. That's worked

27:04

really well for our

27:07

partners like the Scottish Parliament -

27:10

they couldn't deal with it

27:15

if it was just arguments

27:16

all the time.

27:18

But gamification can go a long

27:20

way. We actually have

27:21

an idea we call

27:22

the Community Challenge.

27:24

We actually applied to Google.org,

27:28

though we couldn't get that through.

27:29

But basically it's an idea

27:30

where governments or civil society

27:32

can put forward challenges

27:34

in the city or in a country,

27:35

and then a team of human

27:37

players with AI agents work together

27:40

to solve the problem.

27:43

They take different roles -

27:45

it's like a role-playing game.

27:47

There's a journalist role,

27:49

an investigator role - different roles.

27:51

It's set up like a game,

27:53

but it actually has real

27:54

issues behind it.

27:55

What's going to enable it to work

27:58

is the AI agents.

28:01

The agents are going to be

28:04

able to keep things from going wrong,

28:05

they're going to be able to

28:07

do the complicated things,

28:08

do lots of research,

28:09

process information, and

28:10

do all the fact-checking.

28:11

That's actually an idea we

28:15

have on the table,

28:16

but it would cost like a

28:17

million dollars or something to develop,

28:21

so one day maybe.

28:23

I was thinking related to

28:26

institutions - a platform like

28:28

the one you are

28:29

describing, how easy or

28:31

hard is it to explain

28:32

to an institution, to a politician,

28:33

this innovative platform?

28:34

I don't think it

28:36

is hard to explain in terms

28:38

of how it works.

28:39

It's very simple to use.

28:40

For example, in Iceland

28:42

we have forty thousand

28:44

registered users in Reykjavik,

28:46

but totally ninety thousand

28:48

registered users in the

28:50

whole country because they have

28:52

fifty-plus pilots using it.

28:55

So like one-third of the

28:57

country knows the platform,

29:00

including the politicians.

29:04

Many people have used it.

29:07

That's a bit of a special

29:10

situation here in Iceland

29:12

in terms of how many people

29:15

have used this open source platform

29:17

specifically. But it's really about

29:21

does it serve a need?

29:23

That's what we need to

29:25

convince the politicians about.

29:26

But it's not enough to

29:27

convince the politicians because

29:31

you also have to set it up

29:33

so the layers of

29:35

bureaucracy that need to

29:37

respond to it are going to work

29:40

there. Any policy decision

29:42

involves a very complicated

29:44

interplay between different levels -

29:46

on the city level but also

29:49

on the parliamentary level in

29:51

terms of lawyers, regulations,

29:54

and different levels of

29:55

bureaucracy. So you need to

29:57

have a practical plan

29:59

that's going to make the

32:04

participation work on the bureaucratic

32:06

level. One of the things

32:08

that often kills most

32:11

participation processes is budget -

32:12

budget money. That's the

32:15

reason why we

32:15

started very early on

32:18

with Your Priorities and our policy is

32:20

that if you don't have money,

32:23

you don't need to pay for it.

32:25

It's a volunteer thing

32:26

since we're a nonprofit.

32:29

We've done many of

32:31

the most successful Your Priorities projects

32:33

where they didn't pay anything.

32:37

We got around that "Oh no,

32:39

we have to find money for this" -

32:41

well, it's free.

32:43

That's one thing. For

32:45

most modern politicians -

32:47

smart people working in

32:48

municipalities and governments -

32:50

most of them already

32:52

think it's a good idea

32:54

to get more information

32:55

to make better decisions.

32:57

So it's usually not a lot

32:59

of convincing to do

33:01

or explanation of how it works,

33:03

but it's just everything else -

33:04

all the complications that

33:06

result in

33:06

participation not happening.

33:08

I'd like to share something

33:10

about your organization in

33:12

relation to your team -

33:14

what kind of skills,

33:17

and maybe how was it created?

33:18

Where are you searching for people?

33:21

It started basically -

33:25

it's always been like a

33:25

small group of people.

33:27

It's mostly been like five

33:28

or something in different roles.

33:30

For seven years,

33:32

I took no salaries

33:33

from the foundation.

33:36

We've gotten grants

33:38

from the EU and

33:42

from all over the place,

33:43

but it's always been

33:45

like a passion project.

33:46

So several people have

33:48

come and gone. Recently

33:50

with me, we've had

33:52

Joshua and my partner Robert,

33:54

plus my son Alex

33:55

and some others.

33:56

It's been like a few people around -

33:59

mostly friends and family

34:00

type of thing. When

34:02

you don't have dependable salaries

34:03

to offer, that's usually how you

34:07

get people to help,

34:08

but that is also great

34:10

in many cases.

34:13

Obviously it's been a lot

34:15

on the technology side -

34:17

most of the technological development.

34:19

We partner with, for example,

34:22

Northeastern University Center for Social Impact.

34:25

Beth Noveck has been

34:28

leading those research processes.

34:29

We partnered with her from

34:31

2013 to 2019.

34:33

Also the Scottish Parliament

34:35

and many others.

34:37

We are just bringing the

34:39

technology component.

34:41

That's our domain or

34:42

scope, if you like.

34:43

We don't offer the

34:45

service of going somewhere

34:47

and setting up a democracy project

34:48

for somebody. We have

34:53

partners for that from

34:56

Democratic Society from Brussels

34:59

and in-touch.org,

35:00

European Citizen Action Service.

35:02

That's how we've been -

35:05

lean, family, friends oriented,

35:08

but with a lot of partners

35:11

that help us scale up the work.

35:15

I was curious about

35:18

the projects that you are

35:19

working on or that you worked on.

35:21

If you would like to share something?

35:22

So really exciting project -

35:24

I'm actually going to be presenting

35:27

in San Francisco in

35:29

August at the opening

35:31

with this as part of

35:34

the Future of Work program

35:36

of the GitLab Foundation.

35:38

GitLab is a software

35:40

development security company,

35:41

and they have this foundation.

35:43

We were one of the projects

35:46

that got funded last year.

35:48

This is with Northeastern University

35:51

where we're using AI

35:53

and agents to help states

35:55

identify jobs that are

35:58

requiring university degrees but

36:00

really should be skill-based.

36:02

So this is the idea

36:04

of the problem being that

36:07

there are jobs that

36:09

go unfilled often, and

36:12

maybe some of them have

36:16

too high education requirements.

36:17

So we use AI to help

36:21

states optimize that process.

36:25

This has been a

36:26

really interesting project.

36:29

I think that's where you can see

36:32

in many places this idea

36:35

of using AI for good.

36:38

If nobody is funding it

36:40

and it's not profitable enough

36:43

to have that focus,

36:44

it's really important for society

36:46

to keep funding nonprofit ideas.

36:49

Another interesting project

36:52

from earlier in the year -

36:54

we worked with Northeastern University

36:56

and the state of New Jersey

37:00

where they are

37:04

doing research on the job

37:06

market - what effect will AI

37:08

generally have on the

37:10

job market in New Jersey?

37:12

They have this AI task force

37:16

in New Jersey.

37:18

What we helped them with

37:22

is that we first identified

37:23

the root causes of the problems -

37:26

what are the root causes

37:29

of disruption of AI in New Jersey?

37:31

We did this using our

37:35

open source tools to do

37:38

large-scale web research

37:39

looking at thousands of

37:41

documents, plus interviews,

37:43

to look for everything recent people have been

37:46

talking about regarding the

37:46

effect that AI will have on jobs.

37:48

Then we had a curated

37:50

list of two hundred

37:52

AI impacts or potential

37:54

root causes of problems

37:57

that people might face.

37:59

We put this into voting

38:01

by the public using a tool

38:04

called All Our Ideas,

38:07

which we also developed,

38:09

where we asked people what's

38:12

going to be the greatest impact

38:18

of the next five years

38:23

of generative AI.

38:25

Then people vote in a

38:27

pairwise manner, and we got

38:28

seventy thousand votes.

38:30

Some of the top items

38:32

were things like privacy issues -

38:36

people are concerned about privacy rights.

38:39

Then we did a second stage

38:41

where we actually used

38:43

AI also to help us

38:47

come up with solutions to

38:50

some of the problems.

38:52

It's a very classic example

38:55

of many of the projects

38:58

we've done in the past

39:00

where we are mixing together

39:02

AI research, then

39:04

engagement of people at

39:06

different stages, and then

39:07

we also use AI also

39:10

to help us find solutions.

39:14

Other projects you're excited

39:17

about - could be a project

39:20

you worked on or maybe

39:22

a project of someone else

39:24

that is interesting?

39:27

I think in general there

39:28

are many people

39:31

looking at how to

39:34

use AI for good.

39:35

I think that sort of

39:36

in general, the category

39:38

of projects is that

39:40

the promise of AI

39:41

to do good for society

39:43

is at least as powerful

39:45

as the potential bad things

39:48

AI can do for society.

39:49

The thing is that

39:52

if we don't have the

39:54

imagination to actually think

39:55

about projects to do good,

39:56

then the bad things will happen.

39:58

I think we've missed this

40:00

with social media.

40:01

I was talking about this

40:03

for years - how social media,

40:05

which we saw quite early on,

40:07

had the potential for some

40:10

bad influences on society,

40:12

on mental health and stuff.

40:14

It just went on and on,

40:15

and this idea came

40:16

that we can regulate

40:19

social media with rules.

40:22

Because we can't just

40:24

stop the bad things happening

40:26

with this technology,

40:30

we've been promoting

40:32

and are generally promoting

40:35

that yes, obviously you want

40:37

to stop the bad things,

40:39

but you also have to

40:40

do good things.

40:44

You have to have

40:46

positive ideas and things.

40:47

You can't just try to

40:49

stop the bad because

40:50

you're going to fail quite badly.

40:52

You also have to do positive things.

40:54

I think that's so important

40:58

when it comes to democracy projects -

41:01

we have this amazing opportunity

41:03

now with AI.

41:06

Just like with programming,

41:08

I can do things now

41:10

literally ten times faster

41:12

programming-wise with agents

41:15

than I could two years ago.

41:17

Ten times faster - that

41:19

is like ten of me

41:21

two years ago.

41:23

Think about that.

41:25

If people are having

41:26

interesting ideas for projects

41:29

to do something for democracy,

41:30

the barrier to entry

41:32

is so much lower.

41:35

So maybe a bit of

41:36

a long answer to your question,

41:38

but the projects that

41:39

I'm excited about are

41:42

the new projects that

41:43

listeners are going to be

41:45

working on for this purpose.

41:47

To reply in some way

41:49

to your question,

41:50

I have this feeling that

41:52

software now is not

41:54

so important - I mean,

41:56

it is important, but

41:58

before you had to

41:59

have a team of maybe

42:00

ten developers, and now

42:03

just one person could

42:04

be enough.

42:07

I can feel that

42:08

in some years, maybe

42:10

also my mother who

42:11

is not a developer,

42:13

maybe she will be able

42:14

to talk to ChatGPT

42:15

or Google something

42:17

and create an application.

42:20

So what is important now

42:22

in the civic tech field?

42:24

After it's not so

42:26

important to build a

42:29

platform that is well designed -

42:30

I would say we humans

42:32

should still be the

42:35

designers, or could AI

42:37

also have this role?

42:39

What AI does is

42:41

that it empowers people

42:42

with agency that need

42:44

to want to do something.

42:46

So if you have this

42:48

need to want to do

42:50

something good,

42:50

now it's easier than ever in

42:53

terms of just using

42:55

AI to do things you want

42:57

and empower you and

42:59

a small team,

43:01

one person, two people.

43:03

You're absolutely right - for sure.

43:05

We are the same - we're

43:07

a bit behind the curve

43:10

on programming and stuff,

43:12

but it's getting there soon.

43:14

And that's also like

43:16

with software - software

43:17

is like a means

43:20

to do something.

43:23

The software is just

43:25

there to achieve some

43:27

sort of goal.

43:29

But I think we're

43:31

already seeing that

43:34

in the job market

43:38

where with programmers

43:42

in general, there aren't

43:43

as many openings,

43:47

but there was a lot

43:50

of growth during COVID

43:52

and a lot of people

43:55

put everything into digital solutions.

43:58

Now big companies have

44:02

been laying people off

44:05

like Microsoft let nine

44:08

thousand people go.

44:12

There's this sort of

44:14

interesting situation where

44:16

people are getting let go,

44:19

and you know, "Start

44:22

your own startup" and

44:24

stuff like that.

44:25

That's great if

44:27

they have that drive.

44:29

But I think there's

44:30

also a huge scope

44:32

for people to do

44:33

things in the civic

44:35

tech world and civil

44:38

society that aren't going

44:40

to make big dollars

44:42

but are going to

44:43

make things better for people,

44:45

make democracy better,

44:46

or whatever.

44:49

These can have huge

44:51

rewards if you're going

44:52

to do something like

44:54

that and you're successful

44:55

at it.

44:58

Even if you're not

45:00

successful, it is still

45:02

highly rewarding.

45:04

Obviously pursuing startup

45:06

money and everything -

45:08

that's absolutely fine,

45:09

I've been there myself.

45:12

But I'm just saying

45:14

with this age of

45:16

AI and lower barriers

45:19

to entry, I'm hoping

45:21

also that

45:24

how resources and money

45:28

and everything is distributed

45:31

is going to change.

45:33

But what matters is

45:34

agency and creating real

45:35

value, and there are

45:37

so many things you

45:39

can do with creating

45:40

real value in civic

45:42

tech and with democracy

45:44

that could really make

45:47

people's life better.

45:48

Looking for meaning,

45:50

looking for purpose -

45:52

that's definitely for some people

45:52

going to be a great

45:56

path forward.

45:57

Do you have anything

45:59

you're working on that

46:02

you have some problem

46:03

to fix, so

46:05

you will need some help?

46:06

All the time.

46:08

There are both software

46:11

problems and other problems -

46:13

all sorts of issues.

46:16

One of the great

46:19

things about this part

46:20

of working in the nonprofit

46:22

civic tech world

46:25

is that there are

46:28

often conferences, meetings,

46:30

common projects, and so on

46:31

where you get a sense

46:33

of understanding the human

46:35

aspects of all of it.

46:37

If you're just with

46:39

your head in the code,

46:41

you also need to understand

46:43

the human aspects of it.

46:44

But now

46:47

with AI today,

46:49

if I have a complicated

46:50

programming problem,

46:52

I have several options.

46:53

I sometimes put the

46:54

same question into ChatGPT,

46:56

Gemini, and Claude -

46:58

I call it like

47:02

my AI council.

47:04

Also, ChatGPT's

47:06

OpenAI programming assistant -

47:07

when I get bug

47:09

reports or want a

47:11

new feature, I

47:13

used to put it

47:16

in a task management system,

47:17

but now most of

47:19

them go straight into

47:21

ChatGPT's agent.

47:23

Instead of putting

47:26

it on a list for later,

47:27

I just put it straight

47:28

in and say "fix this,"

47:30

and then it just

47:31

does the feature or

47:34

fixes the bug for me.

47:36

Yeah, we hope

47:38

that continues.

47:40

You were talking before

47:41

about the community of

47:46

civic tech, people

47:48

volunteering or working

47:51

in the civic tech field.

47:53

Do you have any

47:55

message for them?

47:58

Because sometimes there are

48:01

people that have another

48:05

side problem - finding

48:07

funding for their project.

48:09

They really believe in

48:11

an idea that can

48:13

make the world a better place,

48:14

let's say.

48:17

My message is

48:19

that there's been this

48:21

transformation, especially when

48:23

it comes to civic

48:25

tech specifically -

48:26

we have already been

48:29

transformed in terms

48:32

of how much capacity

48:34

increase we can have

48:37

on a small team

48:38

using AI.

48:40

I also think on

48:42

the funding side,

48:44

in the next few years

48:46

when people realize -

48:46

when we see so

48:49

much of our work...

48:51

I consider myself a programmer.

48:54

I've been a programmer

48:56

since 1981.

48:59

I have really worked

49:01

hard at it, practiced

49:03

a lot, become really

49:06

good in my field.

49:07

I've put in

49:09

hundreds of thousands

49:10

of hours of practicing.

49:13

This is what humans

49:15

do - we practice.

49:16

But now I

49:19

have access to

49:21

AI that is almost

49:23

as smart as me.

49:24

The next generation will

49:26

be much smarter.

49:27

That's sort of -

49:28

I think my

49:30

worth and value

49:32

is a bit difficult

49:34

to think about just

49:35

because these neural

49:39

networks are in a

49:42

way taking my job.

49:44

But I think

49:48

the value and what

49:50

you can do in society

49:53

is going to change

49:56

totally when we see

49:57

knowledge-type of work

50:00

being so highly automated.

50:05

There are going to be

50:07

new things we can't

50:10

imagine yet.

50:12

I think funding for

50:15

making society better -

50:17

I think that's definitely

50:18

one of the positive

50:21

potential outcomes we

50:23

can get out of this.

50:24

We basically have

50:26

two scenarios: we have

50:30

a scenario where we

50:32

have like thirty percent

50:33

unemployment and corporations

50:35

control everything, but we

50:37

also potentially have

50:40

a positive scenario where

50:42

we have like a

50:44

three-day work week

50:47

and plenty of resources

50:49

for doing projects

50:51

connected to society

50:53

because there's so much

50:56

value created.

50:58

If you do anything

51:00

that actually helps

51:01

facilitate better things,

51:03

that has value.

51:04

There are so many

51:06

things in society

51:08

that are under-invested

51:10

in because of how

51:12

capitalism works in

51:14

the world,

51:16

mostly in the US.

51:17

This is much better

51:21

in traditional civil

51:23

society being well-funded

51:24

in Europe, not so

51:26

much in Iceland.

51:27

But at least that's

51:29

my belief - that

51:32

we're going to start

51:35

to really look at

51:37

what's going on around this,

51:39

what is important,

51:43

and when we see

51:44

things that used

51:45

to be very expensive -

51:48

like programming -

51:50

plummet to almost zero,

51:51

I think we're going

51:53

to see more clearly

51:56

what is important.

51:58

I think there are

52:00

a lot of opportunities

52:02

for people to

52:04

tackle civic tech,

52:05

democracy, and so on.

52:07

Thank you, Robert.

52:09

Would you like to

52:11

say something more?

52:12

For me it was

52:14

great to thank you

52:17

for having me on here,

52:20

and I'm sure we'll

52:22

speak again.

52:23

Absolutely.

52:25

52:25