Thank you. Thank you for having me. I'm honored to be part of this integration conference

at Zollhof 7. It's another reason to hope that Corona is soon over because then more

people can enjoy this beautiful occasion. Thank you, Katharina, also for that beautiful

insights about some of the most important challenges we have when implementing AI. It's

important to understand what is needed to have an AI make what an AI is supposed to

do. What I try to reflect now in the next minutes is about a complementary part of that.

It's about the relationship between AI and purpose. It's about what purpose do we expect

an AI to have and what the concept of purpose can help in order to make proper AI innovation,

proper AI exploration. So as we already heard before, when we started not only the new fiscal

year for Siemens, but also a new era in many dimensions for Siemens, there was always one

term, very dominant, but that is that Siemens is focusing on technology with purpose. I

dare to say that AI in that peer group of fancy new digital technologies has quite a

prominent role. I wonder why that is. I would say it's basically about three reasons that

are driving that special role of AI in that context. One reason is the term itself, artificial

intelligence. I mean, we as human beings are considered to have our USP because we are

intelligent, right? And then there's artificial intelligence. So this is somehow potentially

competing with our USP, with the crown of the creation. So that opens the ground for

further suspicious thoughts here. And the second reason is that under the umbrella term

of AI, there are so many different capability areas. We have come up with a use case structure,

catalogue of, okay, what actually is all can be found, what is embraced by an AI. And the

third reason certainly is that in each of these capability areas, which we find, there

are very different use cases that can be differentiated when we have a proper look at the purpose

behind these use cases. And just a few glimpse examples, which you probably all know, but

just to understand the concept behind it. If you look at large scale monitoring, using

large amounts of basically visual data to take on machine driven interpretations of

a specific situation, then that capability is used, of course, in an industrial context,

when you talk about monitoring infrastructure. In our case, it's monitoring power lines,

pipelines, rail infrastructure, all that. It's a completely different thing if you use

that very same technology to control whole societies by putting cameras anywhere in the

public space. Another example is a creative AI. Of course, you can use generative adversarial

networks, for example, in the design and engineering process called generative design, an AI coming

up with hypotheses, what designs, what architectures could be possible. And then the expert in

the robots or AI human collaboration teamwork can then decide which to pick. On the other

hand side, you see creative AI producing fake news, fake pictures, fake videos, and thus

already have a societal impact and up to being able to manipulate democracies. And maybe

the bluntest of all examples, of course, autonomous physical systems help to develop the traffic

towards a zero harm situation, right? With lowering mortality rates through autonomous

vehicles, in our case, of course, trams and cars. And the very same technology can, of

course, also contribute to the exact opposite, right? So this very blunt example is what

I want to make clear here, that it is about the purpose behind it. It is about what do

we want to achieve with an AI. And now it's a philosophical question. I don't want to

get into what is good, what does mean AI for the good, for example. But it may be a good

heuristic to use the United Nations Sustainable Development Goals as an approximation here,

right? They have set in 2015 a nice set of 17 SDGs. And of course, here you can analyze,

okay, is there a correlation? Is there a contribution to one of these Sustainability Development

Goals? And if yes, then the AI, of course, contributes somehow in a purposeful way. And

the second layer, this is what, for example, we have heard before from Katharina, responsible

design, trustworthy design. Is the AI doing what it's supposed to do? Is it contributing

in a responsible way? Or are there any negative side effects or any unwanted effects which

we have to take into account and which we have to mitigate and think through upfront,