Latest / The Road to Accountable AI / Jean-Enno Charton: Operationalizing AI Ethics

Transcript

Jean-Enno Charton: Operationalizing AI Ethics · The Road to Accountable AI

0:01
Shaun O 'Noe, thank you so much for joining me on the road to accountable AI.
0:05
Thank you for having me.
0:08
So what is responsible AI or AI ethics?
0:12
Pick the term that makes the most sense to you.
0:14
What does that mean to you?
0:17
So AI ethics, or for us, it's also how we call it, digital ethics under data ethics
0:24
and AI ethics for me is the responsible handling of algorithms and AI.
0:30
So how can we responsibly use and deploy this technology in the world so that key
0:37
ethical principles are safeguarded so that we create trust within our solutions?
0:41
That may be trust from partners, trust from...
0:45
even our own employees plus from customers, that we offer solutions that
0:50
they trust us to buy solutions or work with us.
0:53
That's the whole idea behind this.
0:55
And to do this, actually, even besides regulation or an absence of regulation or
1:00
developer regulation, we want to do the right thing.
1:02
And how do we do this?
1:03
By behaving ethically in our AI solutions.
1:06
That's the not so short answer to responsible AI.
1:11
Before you had a digital ethics program, as I understand it, you had a bioethics
1:15
advisory panel.
1:17
How did that earlier work inform what you're doing on digital ethics and AI?
1:23
So yeah, actually, so with the company, the bioethics work started already in
1:27
2011, which is some time ago.
1:28
And this arose from, let's say, bioethics challenges for us.
1:32
We have a strong health care pillar.
1:33
One of the pillars is health care.
1:34
And it actually has quite some sort of like obviously, we're going to try the
1:38
programs, but also our fertility research, one of the leaders in female fertility.
1:42
Actually, this brought quite some questions around sort of like research on
1:46
embryos, embryonic stem cells with it.
1:49
And how do we deal about this?
1:51
And sort of at that point, 2011, there was quite a big debate around this.
1:55
And so we decided to get external help in the sort of like the Merck Bioethics
2:01
Advisory Panel was then called.
2:02
And they really helped us to come up with principles and structures and how to
2:06
address ethical questions.
2:07
And then when five years ago, we first ventured really into digital health with
2:12
the collaboration that we have with Palantir called Syntropy.
2:16
We really looked into sort of like, okay,
2:18
How do we do this ethically?
2:19
Because as I mentioned, the trust element is really critical for us.
2:24
And then we looked at sort of like, OK, what do we have we learned from the
2:27
bioethics?
2:28
How do we operationalize it?
2:30
How we put this forward?
2:32
And that's sort of like how we borrowed from the bioethical principles and
2:35
frameworks and everything that we had done, and also these advisory panel work
2:39
into digital ethics and then also AI ethics.
2:43
And that's something that I guess why pharmaceutical companies have a little bit
2:47
of a...
2:47
Head start here because we have been dealing with these ethical questions all
2:51
along and we are a highly regulated sector at the same way.
2:54
So that's not, we're not the first, not the first time we encountered this.
2:57
So we can borrow some ideas we've had before, other than maybe let's say big
3:01
tech who have not been in this situation before that there's like also like a very
3:06
critical public having a look at you.
3:09
So like, how do you behave doing this rightly and really being careful that you
3:12
do the right thing in your innovations.
3:15
What's different about AI ethics from bioethics?
3:22
Why not take the existing advisory board that you had and just extend it into this
3:25
area?
3:26
Yeah, so we realized we needed very specific expertise.
3:31
Actually, we went with our first digital ethics questions to this existing advisory
3:36
panel and said, OK, can you advise us how to do this?
3:37
And they said, yes.
3:38
We can give you a few hints, but this is out of our realm.
3:41
This is not something we can help you with.
3:43
We can give you some basic principles.
3:44
But first of all, you actually, for yourself, you need to set yourself, let's
3:48
say, ethical standards or ethical principles that then you can derive your
3:52
tools from.
3:53
Because we always come from the
3:56
bio and medical side, and we have no experience at all in the digital sea.
4:00
I mean, we have some that have sort of like big data, medical, big data
4:03
experience, but the skill set is very different.
4:05
And that's also why our digital ethics advisory panel is, other than the
4:09
bioethics one, is consisted also out of industry experts.
4:12
So they're all external, but they're, let's say, a mixture of academic and
4:16
industry experts, while the bioethics purely academic, because...
4:22
there's hardly enough, let's say, academic insight right now into all the AI.
4:26
And the innovation is happening within the company.
4:28
So it's very difficult to get the right expertise there.
4:32
So we write more on this.
4:34
And then, yes, I said also, for example, our legal experts, they really come now
4:38
from the FDA regulatory medical devices, AI side, to really help us with this.
4:44
And also, the factor of fairness became much more important in this regard.
4:49
So we have an anthropologist, actually, from Johns Hopkins.
4:51
pockets, Khadija Ferryman joining, because he has a special eye on neglected
4:58
populations and how do we really take care that there's not a bias in our systems and
5:03
how can we take care that these people are rightfully addressed also.
5:06
So that's some of the points.
5:08
And then also the principles are not the same.
5:10
I mean, they're similar, but then there's different emphasis.
5:14
And also one thing is that...
5:16
Bioethicists are not too keen on going to principle frameworks and these kind of
5:20
this and the digital ethics are much more open to this because they see at a large
5:24
scale, how can we do this?
5:25
Because within bioethics, it's usually a few larger topics that we discuss.
5:31
And within digital ethics, it's a huge amount of topics that we go into on a
5:34
regular basis.
5:35
I mean, we have more than a thousand analytics use cases, for example, in one
5:39
unit alone.
5:40
And how can we screen this at a high throughput?
5:42
And this will not work without.
5:44
assessments and frameworks and these likes, and will not just by like be a one
5:50
-on -one discussion with an SSS.
5:52
But I will go into this in more detail probably later on.
5:55
absolutely.
5:55
No, it's really interesting.
5:57
How do you find people who have the right kind of expertise?
6:01
It probably was easier when you were doing it, getting started a few years ago, than
6:07
now where every company now feels like they need to bring in people that have AI
6:11
ethics expertise.
6:11
But how do you find people that have the right expertise, but also are a good
6:15
enough fit for your organization?
6:17
Yeah, so I mean, sort of like we had the, we had started a little bit because we
6:21
went through obviously our existing panel where we have to, okay, do you know
6:25
somebody and something's happening?
6:26
But then five years ago, this was a little bit more exotic.
6:30
Nowadays, it's also more difficult because everybody is all of a sudden an AI ethics
6:33
expert.
6:34
So you're really difficult to find this.
6:37
And so when we try to go through this, despite going through our...
6:42
let's say, our esteemed experts that we already have.
6:45
We go through our business that often has already the context to sort of like the
6:49
right people in our digital unit, EMV Digital, that's located in Cambridge,
6:56
Massachusetts.
6:57
They have really good contacts already, let's say, to the academic sphere due to
7:01
this innovation part.
7:02
And so we got somebody along as John Alamka, who works for Mayo Clinic, but was
7:06
like a CIO on Harvard for many years, to give us insights.
7:09
And then from there, we went bit by bit forward looking.
7:12
Also,
7:12
you
7:12
do we need?
7:13
So we scan a little bit the academics here, looking into, OK, what kind of
7:17
academics do we need?
7:18
Is there already something?
7:20
And what kind of expertise do they have?
7:21
Have they published in this field?
7:22
This is obviously very important that we go by publications and see, OK, speakers,
7:27
what's the background?
7:28
And then usually we invite them for a few test sessions and see, does that work for
7:32
panel meetings?
7:33
And through them, we then get exposed to other experts again, also to more industry
7:39
experts.
7:39
So it's the.
7:40
a bit of growing exercise.
7:41
And I think we started with three experts and then one change, then we were only
7:44
two.
7:45
And then by now, I think we're seven experts.
7:47
And what's important to see that we also have a linker between these two panels.
7:50
So there's one bioethicist, Jeremy Sugarman, he's also from Johns Hopkins.
7:55
He's a bioethic professor.
7:56
He has done some work for the Obama administration.
7:58
So he's very well renowned in the US and he's, let's say, a generalist in bioethics
8:03
and he's also sitting on the digital ethics panel.
8:05
So there's a link between them and they can also learn from each other.
8:09
Also share best practices.
8:11
You talked about the distinction between bioethics and digital ethics, although you
8:15
mentioned there are some connections.
8:17
Is AI ethics just a subset of digital ethics?
8:20
for us, it's a subset of digital ethics.
8:24
Because when we started to develop, let's say, ways to ethically handle data,
8:30
because our first business ideas, this started then with, as mentioned, product
8:33
centropy.
8:34
This was about health data and sharing health data and how can we analyze data in
8:38
larger data banks.
8:40
And we decided, OK, we need something for ethical handling of data.
8:44
This was 2018.
8:45
Then AI ethics had popped up.
8:47
but it was not really present.
8:48
But we said from the beginning, OK, we want to do something that encompasses AI
8:54
ethics and data ethics at the same time, because for AI, you will always need good
8:58
sets of data and sets that are handled the data the right way for all the training,
9:02
especially now for generative AI.
9:04
And so we decided to do this under the term digital ethics as one.
9:08
Many might just be also responsible AI, maybe for some the more, especially in the
9:13
IT field, the more.
9:16
understandable term for them, but for us, this is sort of like digital ethics
9:19
encompasses this all.
9:20
So that we really also make sure for data, this is something as a German company or
9:24
German headquartered company, this is also something where there's a high, at least
9:29
within the country, there's a high view on how do we handle data or within the EU in
9:34
general, the GDPR, you have heard about this probably.
9:37
So that's why also we said, okay, we also need a strong footprint on data ethics.
9:42
And so we encompassed this as two together.
9:45
Other than the heightened sensitivity and the regulation, as you mentioned, what's
9:52
different about thinking through these issues in a healthcare and life sciences
9:57
company?
9:59
mean, it's obviously been a heavily regulated environment already, which makes
10:03
it more difficult a little bit to operate.
10:05
And then it's also the most sensitive area you can imagine.
10:08
So the trust element is even more critical for us than for other providers, I'd say.
10:13
I mean, maybe it goes similarly for banks and the sensitivity or insurance.
10:20
But then health is usually the most personal information people have.
10:23
And we see this even in the US, that there's a higher, let's say, alertness
10:29
towards health data than maybe other data being used.
10:33
And so this was really for us the strong push.
10:35
OK, we need to get this forward.
10:37
And also, we're discussing with other companies in the health field.
10:40
And actually, the most reach out from other companies I get is from actually in
10:44
the health care field, because they see they have to be extra alert and extra
10:48
early prepared.
10:49
this because this is something that might help them otherwise in the long run.
10:53
And also if you look at the UAI Act applications within healthcare, we'll most
10:59
likely get in the high risk group and always will need more detailed analysis
11:05
and oversight.
11:08
Cutting the other way though, even within your corporate group, there are many
11:12
different businesses, different units, different products, a range of different
11:16
kinds of issues.
11:17
So how can you, as a large company, you develop one set of guidelines or one
11:22
process around digital ethics that spans all of them?
11:25
so, I we have, I mean, I lead a small team of experts.
11:29
I mean, that's we're the parts directly to the board of directors and we have, we are
11:35
sort of like owning the topic over all our businesses.
11:38
So the healthcare, the life science and our electronics business and also are what
11:42
we call the enabling functions.
11:45
So all our, let's say,
11:46
finance and HR for everything.
11:48
So, and we actually take care of questions for all of them, but we, and we have one.
11:53
of digital ethics for the company that defines our principles.
11:57
And it's not do's and don'ts, it's specifically principles.
12:00
A set of in total 20 grouped under five major principles.
12:03
And from these, we can develop guidelines and do's and don'ts for certain
12:07
applications, certain areas, certain fields.
12:09
So that was very important for us that we do something for the entire group.
12:13
But then obviously, these sectors have very different needs.
12:16
I mean, for our life science sector, for example, for the analytics unit, I already
12:20
mentioned it before, we have one that's called the
12:23
center of excellence.
12:24
We have a platform here on Foundry.
12:26
It's a technology platform.
12:27
And here people can develop analytics use cases from ideation to proof of value to
12:34
industrialization.
12:35
And they, for example, asked us, can you quickly build us a filter so we can
12:39
identify those cases that have ethical challenges?
12:42
And within life science, which is sort of like our business that sells, let's say,
12:47
any kind of lab materials, but also tools like genome editing tools or
12:52
mRNAs and lipids, so also offer services, their data might not be as sensitive as
12:59
they have.
12:59
So we don't think that there's so many use cases that are critical.
13:05
So we just wanted to drive with a filter to find out the views that are critical.
13:09
And so we based on our critical ethics, we develop ethical markers that we then put
13:13
into questions that the use case owners can actually have to answer as part of the
13:16
process in their different stages.
13:19
And we can see, OK, is that ethical risk?
13:21
yes or no, and then we can, if there's an ethical risk or the score is high, then we
13:24
go and analyze it.
13:26
So this is, for example, one way.
13:28
But then, for example, in HR, which is right now one of my closest customers, I'd
13:34
say, with all the Gen .AI solutions that come along, or AI solutions in general,
13:38
whether we insource them or if we develop them ourselves.
13:42
For these, instead of like, these are all.
13:46
Yeah, these are not so many, but then they all have very individual needs.
13:49
So we're currently developing a different framework, a little bit based on what
13:52
we've done before, but much more concrete for HR, which comes with certain
13:56
questions.
13:57
And then we also developed something similar, again, for healthcare, really
14:01
tailored to them.
14:02
So that's really, yeah, the code is a pretty good for all, but then it's like,
14:08
yeah, how do we market our different customers?
14:10
So that's really running a small consultancy here rather than, yeah.
14:15
doing a one -size -fits -all solution, or doing a do and don't for everything in one
14:20
document.
14:26
So for those who are really keen on this, they can read about this.
14:30
There's actually a publication in AI Society about this, how we put this.
14:34
They have published this with a partner from the University of Fittenherdeck in
14:38
Germany, Sarah Becker.
14:39
She has been on our advisory panel also before as one of the leading experts.
14:44
And she's also
14:45
we really went and looked into being a European company.
14:49
We looked into, okay, what kind of principles and frameworks have been
14:52
published in the EU in the last five years back then.
14:54
So this was between 2015 and 2020 when we started this.
14:58
And what has been published towards data ethics and the AI ethics.
15:02
And I think we had like, and may they be from governments, may they be from
15:07
corporations, may they be from...
15:09
NGOs or think tanks, and we found a total of 62 frameworks.
15:13
And this was back then, I guess, now with several hundred.
15:16
And we tried to identify principles and all of them to have a most holistic view.
15:20
And often they were called differently, but we really tried to look into all
15:23
details and to see, OK, is there sort of like one word behind this that we can
15:29
frame with us?
15:30
Because some of them are really marketing speech or semantic differences, but we
15:34
really could narrow them down to the different details.
15:37
And then we...
15:38
After that, we actually grouped them into the five major principles that you see now
15:42
when you access the code on our homepage.
15:44
You can see five different principles.
15:46
And actually four of them are identical with the bioethical overall principles
15:50
that exist.
15:52
And one of them actually is new, is transparency, which is because this is
15:56
actually more or less the key principles when we go to digital ethics.
16:00
This was a larger endeavor that took us more over a year.
16:03
What we also did after is actually we checked whether, okay, how does this work
16:07
within the US?
16:08
principles that have been formulated there, also within Asia and China, for
16:12
example, especially.
16:13
And we see that it actually matches pretty well because it really just defines ethic
16:18
principles in a philosophical manner.
16:19
So it's really, if you would read this as normal employee, it's rather complex and
16:23
we have to easing it down for people to explain, but it really helps us to
16:28
identify challenges.
16:30
And we saw that, for example, we're working well in the US, but also in China.
16:34
And maybe if we look at the Asian...
16:38
geographies, it might more be that the traditional principle of harmony coming
16:42
along, that the autonomy is not as important and there's harmony coming up a
16:46
little bit more.
16:46
Societal harmony is a higher value, but beyond that, we have something that is
16:50
actually quite checked globally also.
16:55
Why is transparency so distinctively important for digital ethics?
17:00
So we think about this for us, the key element or why we do this is really to
17:06
create trust in our solutions that we offer trustful solutions.
17:09
And I think with this high opacity that AI has or the whole big data topic has behind
17:17
it, people, it's really extremely difficult to grasp what is happening and
17:21
everybody knows they're putting the data somewhere on the net and something's
17:24
happening with it, but nobody knows what's really happening.
17:26
And you always imagine these kinds of things that are happening with your data
17:29
or there might be...
17:30
used in a certain way.
17:31
So this is for us that we say, okay, we want to be transparent, we really want to
17:35
share sort of like what we do and how we do it and be as open as possible about our
17:40
sort of like maybe AI solutions and what's happening within them so that people can
17:45
actually trust in these solutions and lose the fear and skepticism of these
17:49
solutions.
17:50
I mean, and this goes not only externally, but also internally with our employees
17:55
here at Merck KGA Darmstadt, Germany, because this is...
18:00
really something that, I mean, obviously we have the whole variety of society
18:04
working for us.
18:05
And also here there's quite a level of mistrust, especially when we look at blue
18:09
color workers, sort of like, okay, the CI may cause my job, but also our white color
18:13
workers are threatened by CHPT and think, okay, what does that mean and what kind of
18:18
solutions are we using?
18:19
And so this is something where we think, okay, we really want to educate here.
18:25
This is where this transparency even plays a very important,
18:29
role in -house.
18:31
So we really educate and be transparent with our own employees so that they
18:37
understand this.
18:40
One of the reasons that I was interested in speaking with you is that you have
18:43
published a couple of these papers, which we'll put in the show notes, talking about
18:47
your approach.
18:48
And I was particularly interested in one that talked about what you call principle
18:52
at risk analysis for operationalizing AI ethics.
18:55
Can you talk a little bit about what that means?
18:58
Yeah, so this comes a little bit.
18:59
So why did we actually put this code to digital ethics?
19:01
I mean, why do we need this?
19:03
And I couldn't we just do with the bioethicists that advised us because we do
19:06
not have like a code of bioethics for the company.
19:09
And this comes to the fact that there is not yet a fixed sort of like set of
19:13
principles for digital ethics out there.
19:15
And bioethics are very well established since 30 years or so, but for us, it's not
19:19
the case.
19:19
And so we realized, okay, we cannot just simply form the digital ethics panel, but
19:24
they need something to orient themselves on.
19:27
And these principles that we formulated in the code, they are as guidance for the
19:33
experts to take decisions, but also, obviously, as an overall guidance for the
19:36
company.
19:36
But also, and this is why it's really designed in this way, that we can build
19:41
analytical tools from it.
19:43
And this is the para, as we call it.
19:47
This tool, actually, we developed directly from the code, which is a way to analyze
19:52
the certain questions and see whether in a certain business model,
19:57
things are at risk or principles at risk.
20:00
And with this outcome of this analysis, we could say, OK, if we have a certain
20:03
question, for example, are we responsible how things are done in a certain way and a
20:07
certain technology?
20:08
And then we can, an expert, not everybody in the company, but our expert, my team
20:12
can do this analysis.
20:14
And then we find out, OK, five principles of our total 20 are at risk.
20:19
And then this for us is the discussion basis then with the panel.
20:23
I mean, that might not be a.
20:25
complete identification of all ethical risks, but it's really a good head start
20:28
because often at the beginning of discussion around an ethical, a certain
20:33
ideology, I think challenges, it's all over the place and you need to structure
20:36
it well.
20:36
And so this really helps us to go sort of like in the risks on certain principles
20:41
and how can we address them bit by bit.
20:43
And so this is also like a bit of a loop.
20:46
So a question comes to us from the business.
20:48
We sort of, if it's a complex question, we run this through this analysis and the
20:52
report then goes to the panel.
20:53
And then they can prepare themselves.
20:55
And then we sit down with a panel and discuss all of these risks and how can we
20:59
mitigate these risks, what has to be put in place.
21:02
And then this goes back to the business as a recommendation.
21:05
And then we see that a loop of reporting several times to see, OK, is there
21:09
actually a solution that's working?
21:11
So this is sort of like, how can we actually operationalize this?
21:15
And this is, let's say, for larger complex questions, how can we do this in a better
21:19
way?
21:20
I mean, if we would run the power on any kind of
21:22
of most questions, then I would need like 20 people in my office.
21:27
But for those couple of discussions, it's a focusing mechanism, it sounds like, for
21:31
the discussion that's an input to the advisory board.
21:33
And also really shows you the risk that might arise and something that you might
21:37
not just not have on your radar that might happen.
21:39
And then the output, as I understand it, of the advisory board, as you said, is a
21:43
recommendation.
21:43
What happens if the business says, we don't agree, or the business benefit of
21:48
this is too great, we're going forward anyway?
21:51
So this is always the problem.
21:53
There are two different approaches to how do you digital ethics.
21:55
I mean, there's the way of ethics -based auditing, which is more through
21:58
compliance.
21:59
So there has been a paper of Yako Mukanda on this with AstraZeneca.
22:03
I think this has been done.
22:05
They showed this value of audit every project.
22:09
And in our company, we run it a bit different.
22:11
This is also an experience with bioethics.
22:13
We are awareness -based.
22:15
So we do trainings, and people know that they exist, that they can address us.
22:20
And so we are sort of like, people are not forced to ask us, but sort of like they
22:25
come to us for questions.
22:27
And throughout the years, or with the bioethics, we have built a very good
22:29
reputation.
22:30
So people come to us because they need the help and they see this is actually helpful
22:34
for their business.
22:35
They could ignore us and they could also ignore the advice, but it's well
22:39
documented within the company and sort of ignore the advice and later on, then run
22:43
into problems.
22:45
Then this actually, I mean, obviously we'll heard.
22:48
very much the decision -takers, let's say, reputation with the company or with the
22:54
company.
22:54
So that's sort of like indirect pressure.
22:57
And it also goes to the way that we say, OK, for us, we're a family -owned
23:00
business, or let's say we're on the stock market, but 70 % is off -stock market and
23:03
family hands.
23:04
There's a higher sense of, let's say, generational thinking.
23:10
So we want to have a sustainable business in 15, 30 years for the next generation of
23:14
the family.
23:15
And so this whole of responsible business has been longer.
23:18
thing, and it has also been, it's on the responsibility is on everybody's hand.
23:25
So it's not just with the data scientists or AI developer that's on the all end, but
23:31
it's also with the middle manager, the high manager and the board of directors.
23:34
So it's really the entire chain.
23:36
And this is sort of like a certain culture we have established with the company.
23:39
And this helps that actually people come to us and they take these recommendations
23:43
seriously.
23:44
I mean, they ask us for certain questions.
23:46
And I must also say some recommendations, obviously, we are not able to implement
23:50
because just they may be not realistic.
23:53
But so we come back and we ask, OK, can we have a more pragmatic solution?
23:56
But do you think about this?
23:57
We approach it in a certain way.
23:58
But sort of like how you think it, it's not working because probably they do not
24:03
have that insights in our processes and business.
24:07
And then we really go iterative way to find a solution that's best fitting.
24:10
And I also have to say that our advisors are rather pragmatic.
24:12
I mean, so we're not going to.
24:14
We don't have ivory tower from an analyst that comes a certain way, but it's really
24:17
OK.
24:18
This is a pragmatic solution.
24:19
And obviously, we cannot aim for total perfection.
24:23
But let's say risk mitigation on a larger level, we do go a pragmatic way.
24:31
What has changed over the time that you've been doing this digital ethics and AI
24:35
ethics work?
24:35
Obviously there's now a lot more adoption of AI, there's generative AI.
24:39
How has that influenced the nature of the work you do or things that you've learned?
24:44
So actually, five years ago, I took over the team of digital ethics or bioethics,
24:49
actually.
24:50
I built this up before it was supposed to side job of the chief medical officer and
24:53
I said, okay, we need somebody skilled for this.
24:55
And I saw this sort of like coming up.
24:58
It was just me as a lone player back then and sort of pushed this little bit forward
25:04
with the company, but it was seen as a rather, let's say, esoteric topic or
25:09
something like data governance or data anthology that nobody wants to do.
25:13
It's something of a side business.
25:16
And I've really seen the, let's say, in the last year, I mean, the people that
25:23
come to me and ask me questions has been risen exponentially within the company.
25:27
I mean, this is Chagy B .T.
25:28
finally showed people things are happening and actually I need to address these
25:32
things and I want to have certain solutions and how can I have them?
25:34
How can I have them responsibly?
25:36
This obviously has also been a sort of like under the attention of the board.
25:39
And we now have a...
25:40
since a few years also a chief data and AI officer who's really taking care of this.
25:44
And he's actually the chair of the panel.
25:46
So he's a very close partner in crime with me.
25:50
He's really pushing this forward.
25:52
And we see a big momentum.
25:53
We have a big data culture team within the company now that's really pushing this
25:58
forward.
25:58
We do a lot of awareness.
26:00
And I mean, I can see that there's not only me pushing and standing in the
26:03
corner, but there's also a big pull effect from different businesses.
26:06
So I don't need to stand on everybody's door anymore and say, OK, I'm
26:09
I'm very sure you have some questions for me, but people are coming running to me
26:13
and ask questions.
26:15
And it's also externally.
26:17
I was just yesterday sitting on a panel also with somebody from Google Cloud
26:23
Service, actually, from Munich.
26:26
And she was also saying that all of a sudden since a year, this whole data
26:30
governance, data ethics thing has moved out of the shadow.
26:33
And she always says to people, this is unsexy in the beginning, and people need
26:37
to get their basics right.
26:38
And it's sort of like,
26:39
eating your veggies every day.
26:41
But this is something you have to do in order to be able to innovate fast.
26:44
And this is also what our chief AI, data AI officer who worked with automotive
26:51
sector before always says, I mean, he always comes to me and always say, so why
26:55
do we have brakes in the car?
26:57
So the brakes we have, so we can go safe, but fast.
27:02
So it's like, okay, this what we have enables us to innovate really rapidly.
27:07
But sort of like in a...
27:09
safe matter so we actually can hit the break if needed.
27:13
One of the issues with generative AI is these foundation models, which get
27:18
developed by big tech companies, OpenAI, Google, and so forth.
27:22
And then an organization like yours might be using that model and bringing in some
27:27
of your own data and building queries and specialized applications, but you're not
27:31
the ones who are controlling the original training of the model.
27:33
Have you had to confront that issue, or how would you think about that in terms of
27:38
your ethics work?
27:39
For example, so we actually use an insourced model within the company that we
27:44
have our own, we have our own GPT, my GPT that we have in the progression where we
27:48
can actually, let's say it's cut off so we can put in our own data or let's say
27:52
confidential data from the company.
27:54
And, but obviously there are questions about, okay, how do we do this?
27:57
And obviously the IP questions and the larger frame sort of like how it was been
28:02
trained, it's becoming much more prominent now.
28:04
So how do we, how do we deal with this?
28:06
But.
28:06
But one question that also came up in this whole thing that in our speak up line,
28:11
actually somebody asked us, so you see, I'm a compliance speaker.
28:15
We got the questions that, okay, there were those reports that the OpenAI had
28:21
workers in Kenya, actually, that sort of like were exposed to violent materials or
28:28
pornography materials and didn't get the right counseling.
28:30
And so they did this for like $60 a day and they were really suffering a lot.
28:34
And how can we...
28:35
what kind of thing is this?
28:37
And so we really looked into, okay, is this a technology we can use?
28:41
Has this been developed under human rights violations?
28:45
And obviously we did a survey, and we see it's not the case.
28:47
And also these kinds of programs have been halted.
28:51
And so OpenEye is no longer pursuing this, so that's why we can use this technology.
28:55
So this is some of the things that are in our radar.
28:59
And I think you can read about this, it was a time article, I think, some time ago
29:02
on this.
29:03
And it's something that we take seriously and survey.
29:07
But obviously, then we try to also when we in -source solutions, especially now
29:13
within HR is a big topic right now.
29:15
And we actually have developed also ways to assess the vendor of a product.
29:20
So in the final, sort of like when we have like two or three finalists in our vendor
29:27
selection processes, actually we come now.
29:30
for AI solutions with the certain kind of questions that we need to check also, not
29:34
only sort of like how does the company behave, what kind of principles they have,
29:38
but also how they address issues, how do they have like bodies, the digital ethics,
29:43
so they have sort of like a crisis model for this or certain things, just do they
29:48
take this topic as seriously as we do.
29:50
So that's also a big thing because most AI solutions, I believe we will insert, I
29:54
mean, we will use the few skilled people on this we have, maybe like 200 or so.
29:58
to develop solutions for research, clinical research, prediction tools, and
30:02
these kind of things.
30:03
We're not going to use our in -house experts to develop HR tools.
30:06
Obviously, these will be mostly in -source and maybe adapt.
30:10
And here we need to find a good way of sort of like, OK, how can we ensure that
30:13
this aligns with our principles and general vision for AI?
30:17
You mentioned earlier on the European AI Act, European Union AI Act, as a company
30:24
that was doing this digital and AI ethics work for a long time before that
30:29
regulation was adopted.
30:30
How do you anticipate that influencing and impacting your work?
30:36
So this was also a case where our work was suddenly highlighted and was seen as very
30:43
valuable.
30:43
A few weeks ago, the consultancy highlighted the survey of sort of like a
30:50
UAI readiness within the company.
30:53
And we saw that we have very strong foundation, thanks to all the work that my
30:56
team has been doing in the last five years.
30:59
So that's a good head start.
31:00
I mean, it's also good for us to see, OK, we're doing something this.
31:02
Because the whole thing was for.
31:05
as I said, how do we behave rightly in the absence of regulation?
31:09
So we had the sort of like future regulation probably upcoming already a
31:12
little bit in mind when we design certain things.
31:15
And so the influence on us is not going to be that strong.
31:18
I think it's a little bit more about building sort of like one of the things
31:23
that we need for now is for to identify high risk cases and really document them
31:27
in the right database.
31:30
And so I have a look at this.
31:31
So this is going to be a little bit of more.
31:32
building a broader AI governance that we do not have in that strict way right away,
31:36
because right now it's more decentralized.
31:38
So to really enable innovation.
31:40
So we will move to a little bit more centralized AI governance that also gives
31:46
like a consolidated mandatory review, I guess, for high risk cases.
31:49
So it was no longer voluntary, as in my case, but then there were probably like a
31:53
legal ethics and whatever other body review of certain projects and areas that
31:59
we identify really in response to this.
32:02
regulation.
32:03
So it may slow down a bit in things.
32:04
And let's say it's not a perfect regulation.
32:07
But I mean, I obviously have two hearts, one of them in the company, I want to be a
32:12
bit more free, but I'm also a citizen.
32:15
And sort of like I as a citizen, I really much appreciate that the European Union is
32:19
moving forward with this.
32:20
It's not an ideal framework, but I think it's still better than having no framework
32:24
at all.
32:25
And you will see how this plays out.
32:27
But I think it's also going to be a little bit of a role similar to the GDPR, which
32:31
we have pushed forward.
32:32
which is lived very differently in European countries.
32:35
So in Estonia, it's lived very differently and much more openly than in Germany.
32:38
So I assume it's similar going to be with the AI.
32:42
But I think it's a good thing moving forward with this.
32:44
And I think also other world regions will look at us pushing this forward and how
32:50
have we done this and what are some of the key elements.
32:51
And obviously, everybody wants access to this market.
32:54
So let's see how this plays out.
32:58
And also how the documentation requirements in the end really are.
33:02
For us as a larger company, obviously it's easier than maybe for like a hundred
33:06
people player.
33:09
Shona, thank you for a really enlightening and instructive conversation.
33:14
Thank you.
33:14
I mean, for all those who see this, I mean, don't hesitate to reach out to me
33:18
through LinkedIn.
33:19
Learn more about how our company is doing this.
33:22
And we're willing to share.
33:23
We actually do a lot of cross -company initiatives also to share best practices,
33:27
pushing things forward, educating, yeah, really policymakers, also exchange with
33:35
academics that do case studies on us.
33:37
So feel free to reach out to me or my team members and discuss.
33:42
Wonderful.