0:02

This is a Technikon podcast.

0:08

Humankind has been talking about ethics

0:10

as long as we have been talking, but

0:13

ethics and technology now, that's

0:15

a relatively new conversation. After

0:18

all, when we think of ethics, technology

0:20

is probably not the first thing that comes

0:22

to mind. I'm Peter Balint

0:24

from Technikon, and I'm happy to bring

0:26

you this special podcast series entitled

0:29

Ethics and Technology - A

0:31

Prerequisite for European Research.

0:34

We can't tackle the entire and often

0:37

fluid topic of ethics, but

0:39

we can show you that things are being done to avoid

0:41

unintended ethical consequences

0:43

and inequities. We can

0:45

explain some of the more common ethical

0:47

issues that become apparent in technology

0:50

projects. And we can tell you

0:52

how these issues are abated.

0:54

And that is exactly what we will do in this

0:56

series. We will look at ethics

0:59

in the context of technology from

1:01

many disciplines, including cyber

1:03

security, infrastructure

1:05

and smart mobility, artificial

1:07

intelligence in personalized health care

1:09

and forensic software and hardware

1:12

engineering, to name a few. We

1:14

will talk about attitudes towards ethics,

1:16

and we will examine how ELSA is increasingly

1:18

becoming part of the framework in EU

1:21

funded projects. And in case

1:23

you don't know, ELSA is

1:25

ethical, legal and societal

1:27

aspects. And if you're running

1:29

a project without regard to ELSA,

1:31

your propensity for complications will

1:33

grow exponentially. To

1:36

kick things off, we delve into the idea

1:38

of ethics in the world of personalized

1:40

medicine, more specifically

1:43

using artificial intelligence to harmonize

1:45

tons of data in order to guide doctors

1:47

towards prescribing personalized treatment

1:50

plans for children with cancer. These

1:52

data consist of things like medical

1:54

publications or large molecular data

1:57

sets. As you

1:59

can imagine, balancing the ethical issues

2:01

with technological advancements is quite

2:03

a task. Today, we are lucky

2:05

enough to speak with Nikola Biller-Andorno ,

2:08

who directs the Institute of Biomedical

2:10

Ethics and History of Medicine at

2:12

the University of Zurich in Switzerland.

2:15

She's an ethics advisor in the EU funded

2:18

iPC project, which stands for

2:20

Individualized Paediatric Cure.

2:23

Welcome, Nicola. Thanks so much for coming

2:25

on.

2:26

You're welcome.

2:27

iPC is a project that relies

2:29

on artificial intelligence to inform

2:32

medical decisions. What

2:35

potential conflicts could arise when

2:37

introducing ethics into a project like

2:39

this and how are these conflicts dealt with?

2:41

In my experience, scientists are quite

2:44

receptive of ethical thinking and

2:46

very willing to comply with ethical standards.

2:48

They just at times tend to underestimate

2:51

the complexity of ethical assessments as compared

2:53

to the science they are doing. The

2:56

High-Level Expert Group on artificial intelligence

2:58

of the European Commission has issued guidance

3:01

that outlines key requirements for trustworthy

3:03

A.I. These requirements include

3:05

human agency and oversight,

3:07

technical robustness and safety, privacy

3:10

and and adequate data governance, transparency,

3:13

diversity, non-discrimination and fairness,

3:15

societal and environmental well-being

3:18

and accountability. So that's a lot

3:20

of stuff to address. And projects

3:22

can fall short. For instance, time

3:24

pressure might lead to the temptation to make

3:26

compromises regarding safety tests or

3:29

commercial interests may tempt developers to posteriorise

3:32

privacy concerns. It's

3:34

therefore helpful if researchers note that

3:37

the ethical standards are part

3:39

of what they expect to adhere to

3:41

and that they are morally accountable for their actions.

3:44

And in highly innovative fields such as

3:46

A.I., scientists have to even go beyond

3:48

compliance and in fact help interpret

3:51

and operationalise ethical principles

3:53

in light of the work they are doing. And

3:56

for me as an ethicist, this is where the fun part

3:58

starts.

3:59

And if we look at perhaps in the

4:01

context of iPC or...

4:04

actually any EU funded project

4:06

for that matter, what gains can

4:08

be realized by adhering to these ethical

4:10

principles other than achievements

4:13

on moral grounds?

4:15

Well, I may be biased, but let me say

4:17

the gains are tremendous. Having a clear

4:19

process for involving ethics early on

4:21

can pave the way towards public trust and

4:23

acceptance, which is what you need if

4:25

you want to sell a product based on your research

4:27

later on. And even more so, involving

4:30

ethics can lead to tools that are indeed trustworthy,

4:32

which is, I guess, what we all want on

4:35

the user side of things. This,

4:37

however, requires a close collaboration between

4:39

ethicists and scientists, which

4:41

presumes a willingness to dedicate resources

4:44

to ethical inquiry that goes beyond

4:46

a token contribution. I

4:48

think such a serious investment in ethics

4:50

is more than worthwhile.

4:52

Yes, and it seems this willingness

4:54

to dedicate resources is overtly

4:57

shared by the EU as they start to

4:59

build in ethics awareness and

5:01

reporting into technology projects.

5:04

Some say the technology is moving

5:06

exponentially faster than ethics.

5:08

This means that ethics can

5:10

never catch up. How should we

5:13

deal with this in the future?

5:14

Yeah, the issue of ethics lagging

5:16

behind scientific innovation has been discussed

5:19

in bioethics for decades, at least

5:21

since the human genome project, when

5:23

people were impressed by the speed with which

5:25

the genome was eventually deciphered. We

5:28

since learned that ethics can contribute

5:30

both from outside and from

5:32

within such big projects. Such

5:35

embedded ethics components are certainly

5:38

helpful to ensure an almost simultaneous

5:40

transfer between scientific discovery

5:43

to ethical deliberation and

5:45

discussing ethical issues together when

5:47

they emerge, or even trying to anticipate

5:50

them, is intellectually appealing

5:52

and provide scientists with an opportunity

5:54

to grapple with their social responsibility.

5:57

Once you've understood and accepted the ethical

5:59

dimension of the work you're doing, it's much

6:01

easier to communicate with the public and

6:03

venture into debates about your own work.

6:06

So it sounds like ethics definitely has a

6:08

place in research and

6:10

technology kind of projects, but

6:13

generally researchers and engineers tend to

6:15

believe what they see; and

6:17

how do we position ethics

6:19

principles among formulas

6:21

and data and complex functions?

6:24

Well, scientists typically have no

6:26

problem at all to understand and use ethical principles.

6:29

Oftentimes, they are the ones who see most clearly

6:31

where questions might be ahead or what consequences

6:34

the practical implementation of a tool may have.

6:36

It's important to encourage such open

6:38

discussions that are uncensured by roles,

6:41

hierarchies of conflicts of interest. Excellent

6:44

scientists are reflective about their own work,

6:46

can be an inspiration to younger colleagues,

6:49

and can help them see that excellence should

6:51

not be limited to technical skills, but

6:53

also extends to ethical deliberation and

6:55

foresight.

6:57

OK, let's go back to this concept of

6:59

artificial intelligence, which is something that

7:01

is being explored in

7:03

your project iPC, Stephen

7:06

Hawking once said that artificial

7:08

intelligence would be either

7:10

the best thing humans ever created

7:13

or the last. So what

7:16

do you make of that?

7:17

I think that's what's so fascinating about

7:19

ethics. We can use technologies in many

7:21

different ways and it's up to us to figure out in

7:23

what direction to take them or where

7:26

the limits, you know, where we

7:28

want to set limits to ourselves. One

7:31

example for this malleability is

7:33

the so-called death algorithm.

7:36

AI is getting quite good at predicting

7:38

a person's remaining lifetime. This

7:40

information can obviously be used

7:42

against people, but it could also be helpful,

7:45

for instance, by making sure they receive palliative

7:47

care when the time has come. If

7:50

we opt to generate this information,

7:53

how we use it, who gets access,

7:55

how it is communicated, all of that

7:57

is up to us humans. So

8:00

I think Immanuel Kant

8:02

was spot on when he said

8:04

in his critique of practical reason:

8:07

"Two things fill the mind with ever new

8:09

and increasing admiration and awe: the

8:12

starry heavens above me and the moral

8:14

law within me." Ethics is

8:16

hot stuff. And I think it will remain so as

8:18

long as humans continue to develop and

8:20

innovate

8:21

And we will be forever innovating,

8:24

which means ethics will always have a place

8:26

in technology. Thank you, Nikola, for

8:28

coming on today and sharing your knowledge with

8:30

us.

8:30

You're very welcome. Thanks.

8:32

Next time around, we look at ethics in

8:35

the realm of infrastructure and

8:37

technology. Efforts

8:40

are currently underway to utilize your phone's

8:42

built in security features to allow

8:44

you to access existing data infrastructures

8:47

to safely enable electronic voting

8:49

and smart mobility. Naturally,

8:52

a project of this magnitude requires a vigilant

8:54

adherence to ethics standards. But

8:57

we dig deeper to see why this project

8:59

goes beyond just data protection. See

9:02

you next time.

9:07

The iPC project has received funding

9:09

from the European Union's Horizon 2020

9:11

Research and Innovation Program under

9:14

grant agreement number 826121