Latest / The In-Between Tech and Trust Podcast / Translating Ethics: Trust, Compliance & the Culture of Responsibility - EP03

Transcript

Translating Ethics: Trust, Compliance & the Culture of Responsibility - EP03 · The In-Between Tech and Trust Podcast

0:07
Welcome to the In Between Trust Podcast, the podcast at the
0:11
intersection of tech and trust. What does it mean to lead with
0:15
responsibility and systems that automate more than we
0:18
understand? Today's guest is Paula Sapir,
0:22
legal expert and ethical AI strategist working at the
0:25
forefront of governance, compliance, and organizational
0:29
culture. In this episode, we explore why
0:32
trust requires both control and care and how compliance can
0:36
become culture, the role of regulation as an enabler, not a
0:41
blocker of innovation, and how language, empathy, and local
0:45
champions help ethical principles stick.
0:49
This is a conversation about alignment, integrity, and what
0:52
it takes to govern AI with human clarity.
0:58
I'm so happy to have your parlor at the in between podcast and I
1:03
would drive dive right into the questions, if you don't mind.
1:05
That was great. Perfect.
1:07
And first of all, it would be great to just get a bit of a
1:11
context of what your paths have like spanned through because
1:15
you've been active in like law, media, public policy, but now
1:19
basically like in the middle of ethical AI.
1:22
And I was wondering like what thread has connected all of that
1:26
for you, like in the topics and also the conversation that
1:29
you've had and what drew you toward the governance of
1:32
technology? Yes, I'd be happy to go through
1:36
that. So I actually have a very non
1:38
traditional background for somebody working in tech because
1:40
my background is actually in the humanities.
1:43
My bachelor's degree was in French literature, European
1:46
cultural studies, and Near Eastern studies.
1:50
Part of that I think is related to the fact that I grew up
1:52
bilingually. My mom is German, my dad is
1:54
British, so I've also always been interested in different
1:58
languages and cultures. But at some point I realized
2:01
that code is actually also just another language that can be
2:03
learned that is increasingly shaping how we live, work, and
2:06
communicate with each other. So ultimately I realized that my
2:10
passion was actually translation, and in particular,
2:13
how to translate legal and ethical principles into
2:16
technical design and day-to-day organizational practice.
2:20
How interesting. And now that like within the
2:22
podcast, we talk mainly about trust.
2:25
And I do think that particularly also in the tech environment and
2:28
the ethics part, it is becoming even more crucial.
2:32
So how would you define trust? Like not from a conceptually,
2:37
but from something that you must build, you must implement and
2:40
you also must feel in terms of like the requirements to protect
2:44
it. And so in German we have the
2:46
saying trust is good, but control is better.
2:49
And trust and control might seem contradictory at first dance,
2:53
right? So if I can control something, I
2:55
don't need to trust it. But trust does not arise from
2:59
thin air. It is based on processes.
3:02
It is based on institutions and the experiences that we have
3:05
around those processes and institutions that we experience
3:08
them as reliable. In fact, I always think about
3:11
pharmacies in this context. So I don't actually know what's
3:14
in an aspirin when I take it. I also only have a very vague
3:17
understanding of how it works. But when I buy aspirin in a
3:20
pharmacy, I trust that what I'm buying is actually working in
3:24
the way that's intended and that I won't poison myself in the
3:27
process when I take it. Now of course, there have been
3:29
plenty of scanners in the Pharmaceutical industry as well
3:31
that I do not want to downplay in any way, but companies have
3:35
been punished for these scandals as well.
3:37
So I believe we will be witnessing something similar in
3:40
the field of AI. And this is something we should
3:42
be striving towards. That we build institutions and
3:44
processes that build trust. That when we purchase and
3:48
interact with AI systems, we can trust that they act and work in
3:51
a way that is reliable. And if they don't, that
3:54
companies will be sanctioned for that behaviour such that
3:56
ultimately, umm trust is not, umm, something insular, but an
4:01
ecosystem that different institutions can contribute
4:03
different aspects. To I fi, I find that so
4:06
fascinating because like the ecosystem part is like something
4:10
that connects topics and also organizations within itself.
4:14
But then again, you also mention of how companies or
4:16
organizations can build up on those foundations within
4:20
themselves as well and form like ecosystems and you but like
4:24
basically at ADDA, you help create like the company policy
4:27
template and also you looked into the AI literacy course that
4:32
is aligned with the EUAI act. And I was just wondering like
4:35
what did you find most challenging about like
4:38
translating those ethical like intense, but also the
4:43
requirements that are needed to form like practical tools?
4:47
So ethical principles can often seem excruciatingly vague.
4:50
What does it actually mean to be fair, to be transparent, to be
4:54
accountable? Oftentimes people struggle to
4:57
translate those principles into their day-to-day work.
5:01
So again, I think it's important for organizations to translate
5:05
ethical principles into something actionable.
5:07
So for instance, that fairness means that similarly situated
5:10
people aren't treated differently.
5:13
Customer relationship management that mine, that might mean ABC
5:17
and recording, this might mean XYZ.
5:19
While it's important that an organization defines legal and
5:23
ethical principles at a leadership level, it's equally
5:27
important to translate those principles into practice,
5:29
practice at a business level. And going through that poses can
5:32
actually be really rewarding and inclusive as an experience.
5:35
So for instance, at some point I worked together with an
5:38
organization that was in the process of developing a company
5:41
policy on Responsibly I. And essentially what they did is
5:44
to bring together different people in the leadership across
5:48
business units who then workshop those principles together.
5:51
In a wild Cafe, a post, there were different stations for each
5:54
ethical principle than as privacy, security.
5:58
And the group would split up and then rotate from one station to
6:02
the next. So at the first station, you
6:03
would define the questions you have about that principle.
6:06
Then you rotate to the next station, you refine the previous
6:09
group's questions, you wrote it to the next station.
6:11
You start answering those questions.
6:13
So step by step, you gain a better understanding of how
6:16
different business leaders might view those principles, what
6:19
questions they might have about them and how they would go about
6:21
answering them. So ultimately what that ends up
6:24
doing is that you develop common frameworks in a common language
6:27
around them that can then help create empathy in an
6:30
organization, but also help that organization develop action of a
6:34
plans around how to implement them into practice.
6:37
So that was a beautiful thing to witness and actually created a
6:39
sense of community in the organization that exceeds the
6:43
sort of like exercise itself as well.
6:45
Fascinating. And I like you just mentioned
6:47
that you found it rewarding. And I was just wondering, like
6:50
what you describe also building up empathy and the common
6:53
understanding, was that the rewarding part or was there
6:56
something else that you like particularly experienced when
7:00
like building those communities of like bringing those people
7:03
together in A room? It was precisely the effect of
7:07
creating a community that transcends boundaries within an
7:11
individual department, which ultimately I think is what you
7:16
need if you are striving to create a culture around
7:20
compliance. So compliance in and of itself
7:23
is anchored in law. You know, there are certain
7:26
regulations that organizations have to abide by and employees
7:29
have to embody those principles somehow.
7:32
But there's often no risk that these compliance exercises
7:36
divide into a kind of like checkbox task.
7:39
You know, every year there's a list of sort of like compliance
7:42
trainings. You have to go through employees
7:43
do them. And then there might be an
7:46
impulse to say, OK, we're done with compliance.
7:48
And I will admit I will be the first one to admit myself.
7:51
Like I work in compliance. I don't enjoy taking compliance
7:53
trainings myself. There's nothing worse that you
7:55
can say. It's like, oh, we all get to do
7:56
like a GDPR training. So I think the the trick is
8:01
really to go beyond compliance. Just think about the kind of
8:06
culture you want to create as an organization.
8:08
What do you want to create as a culture where employees are
8:11
aware that their business is a social actor that has an impact
8:17
on people around them such that they have the desire to assume
8:21
responsibility over the product, say development Apply.
8:24
I particularly love your like view on the culture aspect and
8:28
also combining it with something that might seem rigorous like
8:32
compliance sometimes can be perceived.
8:34
But then again I was wondering what is your take on how like
8:38
the compliance aspect also from a personal or employee
8:42
perspective can differ or needs to be implemented and to
8:48
different be differentiated from true responsibility?
8:51
So oftentimes what I find helpful on the one hand is to
8:55
take a step back and remember that there's a reason why laws
8:59
were developed in the way that they were to begin with.
9:01
So for instance, Germany has very high standards for data
9:07
protection. Oftentimes we complain about
9:10
that at a business level, but we shouldn't forget that there are
9:12
historical reasons why this is such an important value in
9:16
German cultures, having undergone a succession of
9:18
dictatorships that were really exploitative about personal data
9:23
and have done horrible things with that.
9:25
And so it's important to understand their reasons why
9:26
these laws and protections were implemented and why it's
9:30
important to uphold them. At the same time, I think we
9:34
ultimately as as businesses want to develop better products and
9:38
services, right? I think that's what we sell in
9:40
the market. And then I think we framing the
9:42
conversation about compliance such that it's less about
9:45
meeting a legal requirement, that's one aspect of it, but
9:47
more about building better products and services that
9:49
better care to the communities that we're trying to serve.
9:51
I think that's also a way to reframe the conversations
9:54
instead of like doing a compliance training, perhaps
9:56
it's a hack week exercise that essentially strives to create a
10:01
better product that will ultimately lead to a more
10:03
successful company. And I like the business aspects
10:06
are like truly important to also incentivize or also enable, you
10:10
know, to be felt also incorporated into exactly what
10:14
you describe like a bigger picture, also like evolving over
10:18
time. But then again, when you look
10:20
into like other business areas or maybe also academia or
10:24
governance, for example, from your perspective and also
10:27
experience in collaborating in those areas, what are biggest
10:30
gaps right now in like how institutions think about and the
10:35
way that they incorporate compliance?
10:37
Maybe also the accountability of how they use technology within
10:41
the organization, especially also when it comes to the
10:44
activation of AI tools within infrastructure.
10:48
So I think the biggest gap that we confront is often times that
10:52
individual institutions have an imports to act in isolation.
10:55
So we have academia on one end, we have business on the other
10:58
hand, civil society somewhere in between.
11:00
And I strongly believe we need a much more integrated approach.
11:03
They need to be much more open and more regular exchanges
11:05
between academia, civil society, tech and regulators.
11:09
In Europe in particular, what I've often experienced is a kind
11:12
of outright hostility between different institutions, and I
11:17
don't think that's productive at all.
11:18
Ultimately, I think we need to work with and not against each
11:21
other if we are to implement responsible and accountable tech
11:24
solutions in society in a sustainable manner.
11:27
But that means that all parties need to come together and need
11:31
to put themselves into the shoes of the respect of other
11:34
institution, like the kind of incentives and constraints that
11:36
they face and empathize with them.
11:39
In Germany, call it the OM among strategy, the embracing
11:42
strategy, and I oftentimes find that a lot more successful than
11:47
sort of just being confrontational for the sake of
11:49
it. With reference to like
11:52
businesses or intersections within different organizations.
11:56
But then again, there are also, you know, topics that somewhat
11:59
seem differ at some way. And I do also experience that
12:03
there is some friction between like the data protection aspect
12:07
and the enablement of like intelligent systems.
12:11
But how do you see it? Do you see rather like look at
12:13
it from a friction point of view or do you also dive into how to
12:18
create those synergies between both of them?
12:21
So I challenge as core tenets of data protection at several
12:25
levels for sure. First, data protection laws
12:28
generally require organizations to be transparent about data
12:31
processing at the point of collection.
12:34
But a lot of the AI that is currently being deployed already
12:38
was trained on vast amounts of data without individuals
12:42
necessarily having any say on whether they want their data or
12:45
whatever they produced to be used in that way, which in turn
12:49
points a second area of tension, namely data minimization and
12:52
purpose limitation. So AI systems, again, I trained
12:55
on a vast amount of data that was originally collected for a
12:57
different purpose than training AI systems.
12:59
So here's, well, I think we need to find practical solutions to
13:03
bridge that gap. And then finding in terms of
13:05
accountability, there are synergies in the sense that both
13:08
data protection and AI laws strive towards explainability,
13:11
but some AI systems are inherently opaque.
13:14
So then I think it's the responsibility of institutions
13:18
to establish and test in a verifiable way how reliable
13:23
those AI systems are and also implement processes for human
13:26
oversight and responsibility. Because it will ultimately be
13:29
us, the human beings behind the systems, who have to decide and
13:33
also defend ourselves for the choices that we make.
13:36
For instance, which AI systems are appropriate for what
13:38
purpose? What role they played in taking
13:42
decisions that may have a material effect on people's
13:44
lives? So in a way, even though we
13:47
implement more and more intelligent systems into our
13:50
work and day-to-day life, that does not observe us of
13:52
responsibility. Quite to the contrary, it raises
13:55
the bar for human oversight and accountability.
13:57
And that's interesting because you've now said that, like,
14:00
institutions are responsible to, you know, embracing the
14:05
oversight aspect and like 1 angle of it that truly is also
14:08
like discussed widely at the moment is like the regulations
14:12
aspect. And there is like a strong like
14:16
opinion that voices it as like restriction.
14:20
But from your experience, like is it possible to use regulation
14:23
also as a tool for trust and like if So, what would it look
14:28
like to like set up the governance for such an enabler
14:31
and experience it like as an enabler rather than a
14:34
limitation? I absolutely believe that
14:36
regulation can and should be used as a tool to build trust
14:41
and facilitate innovation. And I witnessed this at an
14:43
organizational level every day. If employees do not know how to
14:47
use AI system safely, if they're unsure what kind of data can be
14:50
used and what AI system for what purpose, because they're
14:53
concerned about data protection or information security, they
14:55
will not use AI systems to the extent they theoretically could.
14:59
And there might actually be a disparate impact in there as
15:01
well. So for instance, a couple of
15:03
months ago, a study was referenced in The Economist that
15:07
suggested that women in particular were less likely to
15:11
use AI systems if they fed for instance, that they should be
15:15
doing the work themselves, that they were somehow cheating if
15:17
they're using an AI system or if they weren't sure how to use
15:20
that system appropriately. It's it's called the so-called
15:23
good girl syndrome. So organizations should be aware
15:26
of the fact that if they do not create clear rules around how AI
15:31
systems can be developed and deployed, they might not be used
15:33
to the extent that is possible and that is also beneficial to
15:37
businesses. So clear rules and regulations
15:40
at both the sort of like legal level and the organization level
15:43
can actually facilitate AI adoption.
15:46
I think the biggest issue that we face with AI regulation today
15:49
is not necessarily the legal principles themselves, but that
15:55
the regulator is lagging behind in terms of issuing guidelines
15:59
for implementation. And then we also face a very
16:03
fragmented regulatory landscape in Europe in the sense that we
16:07
have different regulatory bodies issuing guidance around the same
16:11
laws. They might have different
16:12
interpretations of the law. We're already witnessing that,
16:15
for instance, at a data protection level where different
16:17
data protection authorities in Germany have different opinions
16:19
as to whether large language models contain or constitute
16:23
personal data. Uh, this adds to the confusion.
16:27
Um, and so I think, uh, there are certain structural, uh,
16:30
issues that need to be resolved that have less to do with the
16:33
laws than with the way that they're implemented.
16:36
And you've now mentioned, for example, like the good girl
16:39
bias, but then again, you've also touched upon like a bias
16:42
data source. And I do see that as one of a
16:46
friction point within organizations that they don't
16:49
necessarily reflect upon their data governance strategy to
16:52
enable kind of the way that they counter those biases within
16:56
organizations. Is there like an, a best
17:00
practice that you can refer to of how to also counter fire it?
17:05
And if so, so how could leaders learn from it and incorporate it
17:10
into their context of the organization?
17:12
So I think one part of that is to not start with AI and also
17:19
not to see AI as an end in and of itself.
17:21
You actually need to do a lot of cleaning up before you can use
17:26
AI systems effectively. So it's important not to skip
17:30
all the work underlying AI systems that need to be done,
17:33
however, in order for them to work.
17:34
So to be sure that you have an integrated and transparent
17:39
database that is representative for the purposes to which you're
17:42
using it. And we've seen these fallacies
17:45
and big data in the past where we assume that just if we have
17:48
like a lot of data, we don't need to think about causation
17:50
anymore because correlations will speak for themselves.
17:53
And there are a lot of like pretty funny websites out there
17:55
as well that demonstrate how different things correlate that
17:58
are absolutely nothing to do with each other, like the
18:00
suicide rate for the amount of cream cheese that is consumed,
18:04
crazy stuff like that. So it's just important to know
18:07
that AI systems are only as successful as the data with
18:11
which they are fed. So it's important to have
18:13
qualified staff that continually can assess to what extent the
18:17
data is fit for purpose, that has appropriate data science
18:20
training and but that also interacts regularly with
18:23
business and legal teams such that you have these regular
18:25
exchanges that ensure you're taking a holistic approach to AI
18:29
governance and implementation. And we've now like looked a bit
18:33
into the data governance and also structural kind of way.
18:37
And what other kind of questions do you think organizations
18:41
currently should ask themselves before they start or maybe
18:44
before they also enlarge the deployment of AI within their
18:49
structures and workflows? So beyond the technical
18:52
infrastructure, there are also obviously human resources to to
18:56
think about. So we briefly touched upon
18:58
previously that the AIAC now has a requirement of AI literacy in
19:03
the sense that all organizations developing, deploying in the EU
19:06
have to ensure that their staff have a sort of understanding of
19:11
how AI systems work and to be aware of their opportunities
19:15
with some possible harms. So I think one failure mode is
19:19
to select throw AI into the mix and hope it works out.
19:22
It's important to take employees along on that journey and also
19:26
not only address their questions, but also fears that
19:29
they might have around what that might do to their jobs.
19:32
Humans can have a tendency to be very resistant to change,
19:36
particularly if they're unsure how that change will impact
19:39
them. So I think it's important to
19:40
have very transparent conversations about how AI will
19:43
necessarily transform the workplace, but what measures
19:47
organizations are also taking to take those employees along.
19:50
And I think upskilling training, these are like excellent
19:54
personal growth opportunities as well, where the organization's
19:57
goals and sort of like an individual's growth path might
20:00
actually align with each other. The upskilling part, I do think
20:04
is something that a lot of organizations currently like
20:08
look into, but also try to understand of how they can put
20:13
together their employees in like new ways of working and also
20:17
like social frameworks that are required.
20:19
And like one of your strengths, but also the work that you do is
20:23
like translating those complex ethical compliance, regulation
20:27
legal topics into exactly those formats.
20:29
So and what does it take to operationalize like a value in a
20:34
way that it sticks across teams and organizational roles,
20:39
sectors and maybe even languages?
20:41
Yeah, I think the language point is actually the important one.
20:44
It's extremely important to meet people where they are and learn
20:46
to speak their languages in terms of the organizational unit
20:49
that they're in, the discipline, the kinds of incentives with
20:53
which they are grappling to. And this ultimately means first
20:56
and foremost, listening and being empathetic to the
20:58
challenges people and organizations might actually
21:01
face when trying to translate legal and ethical principles
21:03
into practice. There.
21:05
I think it's actually helpful to have local champions.
21:08
So for instance, AI ethics evangelists in each business
21:12
unit that are passionate about this topic that can also then
21:14
help take their their colleagues along.
21:17
Interesting. And how do you identify those
21:19
evangelists and maybe also enable them within organizations
21:23
because I can, you know, both could be challenging for large
21:27
scale organizations. And then again, small scale
21:30
organizations maybe even don't have the resources or capacities
21:33
to provide the room space or even time to look into or serve
21:39
as some sort of accelerator for the whole organization.
21:43
Yes, I think it's important to create spaces where employees
21:47
can come together and discuss the kind of like concerns or
21:52
questions they have in an open manner.
21:53
Such create open forums have also brown bag launches around
21:57
different aspects of the ethics of AI.
21:58
And then I think you'll realize pretty quickly that some people
22:01
show up regularly not particularly passionate about
22:04
them. Then it's important to enable
22:07
those people to continue living out those passions because
22:09
that's exactly what you want in an organization's intrinsically
22:12
motivated employees that care about driving the organization
22:16
forward, but also doing the right thing.
22:19
And if you enable those people to thrive, you will also then
22:23
take other colleagues and employees along.
22:25
That's, I think like a core aspect of building this culture
22:28
around responsibility that I think is so incredibly important
22:32
and, and the age of AI that we're entering.
22:34
I couldn't agree more. And I do think that one of the
22:37
central aspects that organizations need to provide us
22:39
like trust into those capabilities.
22:42
So those persons to be like the accelerator or to also enable
22:47
them in a way that they can serve in their role also in
22:51
their how they speak with the like the organizational context,
22:55
as you said, like from a language point of view.
22:57
But what other kind of like roles or capabilities do you see
23:01
that we need in the future when we look into like building trust
23:05
with an organization? So maybe also building trust
23:08
into particular roles while they expand their knowledge into
23:13
teams and organizations or maybe even partnerships and beyond.
23:17
So AI governance is certainly a growing field and essentially
23:20
requires a range of business, technical, legal and ethical
23:24
skills. So it's a daunting field, but
23:26
also very exciting. And I think implicit in that is
23:29
also the notion that interdisciplinary thinking will
23:33
become increasingly important at the same time, one skill set
23:37
will never grow old, and in fact might become more and more
23:39
important in this sort of like AI enabled environment.
23:43
And that's the ability to communicate clearly and to
23:46
empathize. And actually, I think AI might
23:49
be able to help with that. So for instance, prompt
23:52
engineering is becoming an increasingly important skill,
23:54
which essentially means the ability to give an AI system
23:58
clear instructions, to communicate with an AI system in
24:01
a way that it understands what you're trying to get it to do.
24:04
But of course, the ability to communicate clearly and to
24:07
articulate what you want and what you need is also a core
24:10
human need and skills. So perhaps, and that's my hope,
24:15
the fact that we are now forced to articulate ourselves clearly
24:19
when speaking to AI might also have the positive effect that we
24:22
can articulate ourselves more clearly when talking to other
24:25
humans. So perhaps AI can sharpen some
24:28
of the human skills we need as well.
24:31
I love that because it also could serve as like an enabler
24:35
for a new kind of relationship, right?
24:37
Because you can look into the way that we also as a Society of
24:41
human beings form connection in itself.
24:46
And I do wonder and would be curious to learn from you, what
24:50
kind of relationships do you see evolving now and especially over
24:55
also over time and when we look at us as like a society, but
25:00
maybe after beyond run that and look across institutional
25:04
organizations. So the philosopher and media
25:08
theorist Marshall Mcluhan actually already posited in the
25:10
1960s that technologies were extensions of the South, and in
25:15
an age of both mobile devices and AI, this theory will
25:19
probably be taken to the extreme.
25:21
On the other hand, I think those sort of human computer systems
25:26
might actually actually also help raise awareness that we
25:29
live in ecosystems more broadly that transcend human bodies,
25:34
which hopefully also creates A renewed appreciation of our
25:37
environment too. Because one ethical principle
25:41
and imperative that is currently not featured in the public
25:44
conversation around AI ethics enough is sustainability and
25:47
protecting our environment, particularly considering how
25:49
energy intensive some AI systems can be.
25:53
So that might be actually one of the most important guiding
25:57
principles for the future. How we can ensure that the
26:00
environment in which we live can actually like years, what we
26:06
need to continue living in the kind of way that we want to
26:10
going forward, which in turn can also spur a lot of innovation in
26:13
terms of more energy efficiently.
26:16
I I'm optimistic in the sense that I strongly believe that
26:19
humans will always find good solutions if they are confronted
26:23
with the problem with that obviously also requires raising
26:26
awareness that there is a problem now.
26:28
Diving into the intersection of like sustainability and AI could
26:32
be a whole nother episode. But as we're coming to a close
26:35
with this one, I do want to put it on hold and hopefully have
26:40
you another time with us and wanted to close with some in
26:44
between moments like something that you could answer in like a
26:47
short sentence or maybe just also a feeling that you had.
26:51
In particular, when was a moment when you saw trust break and
26:55
something important in March for yourself?
26:58
Is there something that you could see or remember that
27:01
definitely also required that? Many, many years ago feels like
27:06
an eternity go. In 2013, I was in the middle of
27:09
writing my master's thesis, and I wrote my master's thesis about
27:13
transatlantic privacy governance.
27:15
And then all of a sudden in the middle of that process, the
27:18
Snowden revelations broke, which obviously significantly
27:21
undermined trust in the transatlantic relationships and
27:24
also concerns about the relationship between tech
27:27
companies and the states. This in turn drove the impetus
27:31
behind some of the most significant laws that shape
27:34
digital societies today, most notably the EU General Data
27:37
Protection Regulation. But at the same time, laws need
27:40
to be tailored to purpose. So we are currently witnessing
27:43
large movements towards deregulation on either side of
27:46
the Atlantic because laws are like the effectiveness of those
27:49
laws themselves are being questioned.
27:52
That also demonstrates that that laws and ends themselves that
27:55
they might need to revive as well, but that we need to make
27:58
sure that they address real ethical challenges nimbly and
28:00
appropriately, such that business as a society have
28:04
continued to have incentives to to abide by them and and prevent
28:06
the kind of scandals that we then saw in the past.
28:09
Thank you for sharing. I do think that I can see that
28:12
this was like a point where you were diving into also the
28:17
importance of like realization. Another question, short answer,
28:23
is there like a legal or ethical principle that keeps you
28:26
grounded? And generally I embrace explain
28:29
ability. So both in my sort of like
28:31
personal and professional life, I try to act in a way that I can
28:34
then after the fact, understandably explain why I
28:39
reached a certain conclusion. And even if somebody might not
28:42
agree with that conclusion, I had a reasonably good argument
28:46
behind. That doesn't always work that
28:48
way, but that's something that I strive towards for sure.
28:51
Love it and say I believe that you used to hold but no longer
28:55
do given like the acceleration of our time.
28:58
So I used to actually believe myself that business is somewhat
29:02
operate separately or even an opposition to society.
29:04
So I grew up in a very left wing liberal household.
29:08
But I realized that myself transitioning from academia into
29:12
a business environment, that businesses are ultimately also
29:14
just made-up of people, right? And give people meaning as well
29:18
by virtue of the kind of work they perform.
29:21
Hence, again, it's my strong belief that we need to work with
29:25
businesses and the people that make up those businesses and not
29:27
against them to ensure everybody's aligned towards
29:31
developing and deploying AI responsibly.
29:33
Amazing. And lastly, is there a hope that
29:36
you hold for how AI could governed be governed with care?
29:41
So ultimately it's in all of our interests that we can benefit
29:46
from innovation while at the same time minimizing harm.
29:49
So I am an optimist at heart. I, I strongly believe in the
29:52
intrinsic goodness of people. And if we enable people to do
29:56
the right thing, the vast majority of people will do so.
29:59
So of this I am convinced. And that also makes me hopeful
30:02
that we will find institutions and processes that people will
30:06
follow to ensure that we can develop and deploy AI
30:10
responsibly and sustainably. And to close this like session
30:14
and also like a truly inspiring conversation with you.
30:17
Thank you for sharing all your experience with us.
30:20
I wanted to know if like if trust is something that we
30:24
practice and not promise, what does it mean to be a trust
30:27
builder in a time of automated systems and accelerating
30:32
complexity? Is there anything that you could
30:34
give us with that we can take with us and also incorporate
30:39
into our daily life? It's simple in theory, but
30:43
difficult in factors to do, and that's to lead by example.
30:48
Ultimately, people follow people.
30:49
So if you, as a leader in your business, act with integrity and
30:53
assume responsibility yourself, people will follow on.
30:57
Everything else will fall into place.
30:58
So I think starting with ourselves and to see how we can
31:03
embody the principles that we're trying to convey, that's the
31:06
most important first step. And again, the West will fall
31:09
into place. I think it's the best way to
31:11
close this conversation. Thank you so much for being at
31:14
the end between Trust Podcast, Paula, and for all the insights
31:19
that you gave with us. And I look forward to having you
31:21
at some other point again. I would love that.
31:23
Thank you so much for having me.