Don't look at your phones. Don't look at your devices now because we will have one of the best panels of the day on this stage right now. You chose the right session to come to because we are starting strong with a panel on a challenge that many organizations are facing right now. namely how to move from experimentation to real transformation and from transformation to then the real competitive advantage. So in a moment we will uh invite our three panelists who will be talking about scaling AI from transformation to the competitive advantage and please give a recharged after lunch break uh hand to our three panelists which are Kenza ityabu from scalers Claudia Polling from Figer and Dr.
Oliva from Zmens. [cheering] Thank you. >> That sounds about right. That sounds about right. >> Everybody hope hope you had some food because how active you are. That's that sounds marvelous. >> Pay for you. >> Uh so super happy to to have now talking about AI and scale um as a topic and super happy to have the best panelists ever in the world. uh you know I'm not too sure our our history I don't know feels feels I don't know we've seen each other quite some while now but our history feels very connected to Berlin actually right and and prior to tea times as well so thanks Veronica and and Fabian thinking of us and like let's bring the three together here along the line so topic is and excuse me 10 years later I gained some weight and I need some glasses right so getting aged in here.
So topic is about scaling AI. Uh here we go. Um and two two of the best panelists, the architect the adviser um joining us in order to to to dive more insights. No slides only insights today. And it's a pleasure to have Claudia actually with us. 20 years of battle scars from Tabs Deutschean. She's now director of data and AI at Figer where she spearheads the central data and AI agenda. She doesn't just talk about AI. She builds the water pipes, let's think, on mashes and governancing. We'll talk on that. That keeps the German industrial engine. I love that. Running. If data is the new oil, Claudia is the one building the refinery. How cool is that?
Give it up for Claudia. And we have Kenser, a Spiegel bestseller outdoor. Oh, that sounds so cool. Cancer, that sounds really cool. Former CTO and board member of Figer. I'm not affiliated to Figer um must I say. Now founder and CEO of Scalers. Cancer is the expert who shapes and guides Robert speak and board speak from China to Spain. She's seen it all and her mission is simple. Scaling AI so it actually works for humans and businesses. She's the voice of strategic AI adoption and bridges between complex tech and boardroom leadership. Can say super cool to have you. >> Thank you. [cheering] So scaling in in a some kind of aspect. So last year Andre Kapathy probably the one or the other knows about I said like hey language is the new new program language right um in terms of English is the new programming l speaking on you know wipe coding and prompting right so um that means that that that's something scaled along the line under the predictable wipe let's do a wipe
check in right so kenza if if everyone is now a wipe coder who considers themselves a wipe coder here hands up or almost everyone one is a W coder. All right. Um so are we entering a golden age of innovation at scale or um are we just generating massive amount of spaghetti data which we know all Claudia need to detangle it. What is what is a bit your view on you know is wipe coding a test if you talk to customers to partners or if you advise them are they already embracing that? Is this a holy grail forme industries or the bill industries? Hm. Um I think it's a complex it's a complex topic. Let's see.
There's a difference difference between the motivation and the reality. >> Uh I think vibe coding is really brilliant for experimentation. >> Uh but when it comes to ownership, it's a disaster. So you still need to know what you're doing and vibe coding will help you get there quicker. Um, but if it's not your area of expertise, you know, you believe you're doing the right thing. >> Uh, but you're not. And you still need to explain along the way. Especially in an industrial environment. >> You cannot vibe code a solution and then you don't know what's happening at 2:00 a.m. and then I don't know the warehouse is there's a blackout and nobody can answer that question because it was voded.
I'm exagger exaggerating a little bit but um so my point is is it's is both >> so there's a lot of uh motivation to use it and it's good for experimentation uh at the same time for enterprise use regulated industries everywhere where you need an observability and explainability it's you you should you should be careful how okay >> you use it >> but then in in terms of what white coding made is that accessibility ilities. So everyone is using in certain kinds of copulates assistance. Is there any productivity number one productivity hack you have you know embraced as yourself in using AI that everybody can steal now? Uh I think it's not it's not really a secret.
I do I do use AI and it's definitely helpful because I just started my my own company and I have to take care of everything. So obviously with AI I'm using cloud cloud most of the time. uh you can do many things quicker and if you have agents then uh you have an autonomous um activity delivery let's say at the same time I do very um customized strategical work >> so my brains need still to be awake uh and in those discussions with with the clients so it's it's more that back office uh uh stuff um that I use and I I don't have a huge team anymore this is the bad thing about starting a business >> loneliness.
Is that a hack? >> No, no. AI workers are now artificial, right? >> Yeah. Yeah. Yeah. Exactly. Exactly. So, in in my previous jobs, I could delegate to the team. I had huge teams and I say, "Okay, you do this, you do that, >> and now um and now you have to do everything by yourself with the help with the help of AI." Yeah. >> Claudia, how about you? So we moved from from the wow momentum of 2022 to the how momentum in 2025. Any productivity stack? How are you embedding AI in your daily work and your routines? >> Yeah. So um I don't like speaking about hacks because um there is this wow moment you have to feel.
Yeah. So or this aha moment because um let's just look back the last two months. So we're so crazy. Yeah. So when um anthropic or clo published the sonnet 4.6 my entire bubble went crazy. Yeah. So uh you have to talk in your network about everything what is now possible. So and I had to feel the same. So that's why I tested uh all my LinkedIn data. So I downloaded all my LinkedIn data from 15 years by the way. 15 years. It took it took one day by the way just to let you know. and and then I feed the AI with my data and then I started the the discussion or the dialogue with my data.
So I chatted with my own data and that was really interesting because if you in the past you you searched for a specific topic what you posted three years ago and it was horrible. Now you could just ask so when did I post this and that and what was the conclusion and how many comments and how many interaction that's was really my moment where I thought okay uh everybody has to feel this this potential what's now possible and everybody and I mean really everybody has to test this because it's uh this is it should be tangible yeah so you you you just understand it what is possible when you felt the same aha or wow moment like I did.
>> So in seammens we call this everyday AI and so we're rolling out these kinds of capabilities so everybody can explore somehow you know the GPTs alike called seams GPT with currently more than 190 actually thousand users on the CP GPT for exploring the safe capability and enhancing the productivity. So scale is still the aspect. So cancer what is scale? Can you help us detangling what what does it mean? A let's AI scale. What is this? What does it mean? >> Yes. A CEO from scalers. >> Scalers. This is and this is actually where the name of the company came from. Uh because uh I um observed that most companies are scaling their use case lists.
>> Mh. >> Uh right. and being very proud of oh we have 500 use cases uh but it's like use case ideas on when you ask how many of them have been implemented and how many of them have you scaled then obviously from 500 you have two um and that is the the the question that I am trying to solve right and for me scaling means that you adopted AI in a way that it is part of your corporate strategy >> and it is changing your workflows, how your company makes money basically >> and um if you want to grow as a company, you have to work on your top line. Um using AI might might help but it won't be successful unless that scale isn't there.
Mhm. >> You need the foundations need to be there. The infrastructure, the technology, the data, the architecture starting with the architecture. You need a modular scalable architecture. >> Your data uh management needs to be in place. Governance uh the the operating model. You have to have a look at your processes at your frameworks. And then the most important part the culture and with everything it includes from leadership, change readiness, psychological safety, uh dealing with with failures in the past in in the company. How is the organization open to learning and to continuously learn? >> So this is why it's at the end. So scale for me is is driving that transformation successfully and then the use cases you build.
So wait When you have done your homework, it doesn't matter which use case you choose is going to scale. >> Mhm. >> So end to end along your value chain you because you have access to all the data, your systems are accessible and and and uh there's a backend integration all of these things and if you don't have that then none of the use cases will scale. M >> so basically it's less the question of which use case but more the question of which maturity do you have and depending on that you should use the use cases this is how I understand scale so one one challenge is that that means we you clean your data cla right clean the data and make all perfect data um and then you can scale but in in most of the cases I must admit right it's an a 10 years 10 years endeavor then right how do How do you cope with this?
What is the recommendation? What is the right balance in getting the infrastructure data right or not? >> If you want to clean the data, then you already have the access to the data. But let's be very honest. So this is the theory. Here's the reality. >> Okay. Okay. >> We still integrate there. >> Battle battle arising. Do you hear that? >> No. Can know that very well. This is how I developed my theory because what I have done in the last 20 years was exactly suffering reality. >> But go ahead. >> Yeah. And and the thing is um you have to um start with the data integration and you have to be very sure that you have the overview of all your relevant data and that means also you have to think about relevance.
Yeah. So um it's not it's not possible to connect all the data. That's not the the goal. Yeah, it should be the relevant data because it is of course use case driven. Uh and if if you want to um let's say scale the use case then it's necessary to know which is the right data and which is the the right. So it's it's not about just connecting the data. Then you have the raw material but then you have to uh do a lot more with the data. you have to prepare the data for let's say a dashboard or for the case for the use case and to let's say measure the data quality and and so on.
So that's why if you want to have a healthy data organization there are two possibilities. >> Uh the one is I learned from dojaban um you build three houses. Uh so and that was really genius. Um unfortunately doaban um terminated everything around that but this is real a nice pattern. Yeah. So you need three things. Everything around data foundation. >> Mhm. >> So data governance, data management and so on. The second one is everything around AI where you start with use cases and enable the first AI best practices what you can scale into the organization. And the third one is 3C. We we called it 3C and 3C stands for communication, community and let me think um culture probably no >> communication community and change >> ah there you go >> yeah because uh when it comes to AI you need also change management and this together is a solid foundation this is one opportunity and another one which we trying to solve at at Figer is um to differentiate between foundational work and the
solutional work yeah and then uh you have all everything around data platform, data quality, master data management, uh data governance and also AI governance and self-service because um if you want to scale then you also need to build up self-services. Yeah, because we started from all the reporting stuff and now we want to go in the next uh century which means AI and therefore we have to make space for new things like AI and that's why we put all our data self-services uh in the organization. Yeah. So we enable the organization to build dashboards on their own. Um and the other thing is everything around new things, innovations around data and AI and this is also opportunity or a possibility to to build your data organization.
um this is the reality and it's not ready yet and we need a lot of resources and time and so on and of course patience to put you out there but something interesting is again right it was a bit of an intro ready Europe we're catching that up we're not speed enough we're not scaling enough right and though we had an interesting moment this somehow last year right with open claw Peter Steinberers right coming from jetput November 22 I had the feeling that in 2025 It's always November. Why is it always November that something is released? Um moment seems have changed quite significantly, right? The open source aspect, everyone picked it up and Jensen jumps on on the stage and like oh that's mind-blowing gamechanging.
Would you say that this would be expected? Everybody says like, oh Europe, you're not having AI, you're not making infrastructure, you're not scaling, and suddenly boom, here we have a little rockstar from Austria, single pony, right? Uh, but it's super cool, right? >> Let me answer that because I met a really smart >> person who said to me, and by the way, that was you. >> Oh, >> I I didn't know that >> some years ago. So, um, I I was working for Dodge Telecom and I invited you to our Tabs open and there you said this remarkable sentence to me. you said uh we do have talents, we do have the knowledge, we do have the infrastructure and so on and so on, but we do not have the right marketing, >> we are not good at marketing in Germany.
And that was so true and I always quote you by saying that we should be much more prouder of all our achievements in Germany and Europe. Parloa. Yeah. Right. >> Yeah. So look look we we do have deep we do have Parloa we do have black forest labs and so on and so on. Yeah and we could also be very proud of SAP and also ALF alpha because that are achievements. >> Yeah and if >> and yet we had we have to have some controversy here. um if the German enterprises don't use those solutions then obviously they won't be able to scale and to grow >> because at the end what uh um um um what is relevant for the economy is is the revenues that you make >> and the uh the stock uh uh price that you uh that you generate and and grow.
And this is something where we need to improve. Since a couple of months or maybe two years, everybody's talking about sovereignity and and tech sovereignity. At the same time, only few enterprises are willing to use European solutions. So on the one hand side, we are complaining there are no European solutions. Uh there are some, but nobody's willing to use them. And I remember many years ago when we had the open telecom cloud, >> it was really a pain in the ass to sell that and it was a sovereign solution >> where everybody said okay we need it at least for regulated industries and public and etc. very difficult to sell. Now we're having that moment again and I hope that um uh we will see a shift in behavior there.
>> Yeah. >> What is what is missing? So you know here is here's open claw right and though >> months later open air comes opens the checkbook let's say or a bit of something and environments conditions that matters to those talents right and said like yeah I'm ready to scale this right so why why are European companies not say like I'll do that right this momentum >> mindset Um I'm not sure if it's all European problem. Let's maybe let's focus on on Germany or German speaking countries. D >> um we don't have that failure culture. It's really hard to find managers who have the courage to um work with new solutions. Mhm. >> And even our public institutions, right?
They should be the one actually um trying and and and and encouraging startups and encouraging Germany build businesses. >> Uh but it's almost impossible to work with the the government. You need always references. So if you don't have those references, you're not working with them. And then with the enterprise, they also expect references from from public entities. I think it's a courage thing actually to say okay maybe Al if Alpha is not there yet uh but we want to try a German solution and we will invest with them and and take the time at the end which enterprise German enterprise is really ready to use Gentic AI which one has the maturity to use it at scale almost none so it doesn't really matter does the the Alf alpha or Mistral don't have all features that's open AI I offer because even that's mature enough to use them.
Anyway, so right now at the same time we are talking in a neighbor room there is a talk about Sophie. So we do have a German solution but we do not know yet. So please take care of Sophie with double O. >> But not now. Not now, right? You stay here, right? >> But maybe you will meet Nicholas Harris. Um he is responsible for that I think from FRHA. So uh there there are some developments but we you're right we should use it. Then there was uh you know this is now scaling capabilities that we have in Europe and we are somehow not leveraging and others are scaling that then we have the other side the scaling the business right when when claude came on under the hood and then in combination with open claw um then suddenly you know all the sus business including SAPs you know stock market you know we have a entropic which has valuation I don't know 300 billion something around and destroys trust with the press announcement 300 billion
in talk right uh along the line. So is this the end? Is aentic business the end and we we reshuffleling on that scale the business economics on sus business scales. >> I think it's rather the economic system is at the end. I mean earlier the stock price uh was the proof of the performance of a company. Nowadays, it's not it's totally independent from the performance of of the company. It's just so if we're talking about hype, this is the hype. This is where the hype happens in the capital markets about AI. >> Um I think at the enterprise there is no hype. We're we're just doing the homework still. >> Uh so the capital markets and how it reacts to those news.
>> I think it's it's not realistic anymore. It's it's just something happening in the cloud and it's it's hard to predict anything. >> It's hard to follow. Um >> Yeah. So you think it's more predictable? >> Yeah. Let's look further ahead. So um um agents or we are riding this agent wave right now. Yeah. So um but agents are the same like use cases or data products. It's just a new term. Well, so now you have the list of agents. Um yeah, in the past you had the list of use cases, the list of data product and then now you have the list of agents. Um but when it comes to agents, um then it's just the silo again like employees are.
Yeah. So that's why let's think ahead. Uh what is what is the constant here? And the constant here is um that we access specific um data or specific processes and this is the foundation you should do right now. So uh then you will be prepared for everything what is coming in the next years. Yeah. because master data and processes your your core process that needs needs to be digitized and that needs to be translated into markdown language. Yeah. So this is the foundational work you have to do right now and then doesn't u matter if it's now an agent or multi- aent system or the next wave which could be the AI as an operating system where you access everything around your knowledge.
So now you have to digitize your knowledge in in markdown files. This is what I would do or what I am doing already with Figo. >> Okay. Yeah. >> So that means um in terms of still focusing on you know there's a buzz hype of economic labors but in the end you know >> focus on your >> focus as a best practice focus on your data make the data products right that's the new buzz as well right. So in data products >> not really not really because a data product is already in the past. Yeah that was alum also workaround where you try to measure also value. >> What is a data product just for those who are not not in there >> for me right now a data product is a dashboard a reporting that is a data product and maybe also a standard data product which can be used um more often.
Yeah. So that is a standard that's a data product for me. Do you have another definition? >> No. >> So in if this beta product that maybe has also a data life cycle behind it, right? It's on real time, you know, being provided. Uh they cost money. How do you how do you make make that scale in your organization? How do you get that actually this like okay this is not a that dashboard at the other thousand dashboards that we have but it's actually a used dashboard maybe a paid dashboard or how what is what what is what are the >> I transfer the responsibility to the department which is using this dashboard or report.
Yeah. So we we support with developing that and we of course um deploy the first version or maybe also the second version of that product but then I would like to hand over it to the business and if it's needed they will take it and they will pay for it. If they if it's not needed then you can terminate the product. That's the reality and this is also an kind of mindset and courage because you have to terminate also the data product which you build because it's like I said yeah so uh you have all these lists of developments but this is not only about develop something and deploy you have to run things and that costs money >> and who has to pay for that you cannot do that as a data and the I department for the la next 10 years it's not possible that's why I I would like to hand over these products to the people who are using the product.
>> K. So, do we need cos in organizations? >> Chief chief data officers >> or anyone now a chief data officer? >> Nobody is a chief data officer. That's actually the problem. Yeah. I mean in the in the and it's you're one of the survival surviving cos. Yeah. I mean look look around. most most companies have actually fired or renamed their their co in and something else and uh that that is a problem because understanding that data management and quality are important needs to be um everywhere in the organization and a co per se is the person who orchestrates this exercise and who takes care that this is happening Right. And uh uh that the organization is is taking care of it.
>> At the same time, as a a board member or or investor, you don't see the value right away. >> Mhm. >> So this is why most companies are right now, you know, firing the cos because it's an investment. You said your your data products, you want to to uh um overgive them to the business, right? But the there has been an investment for you to build those data products and it takes a couple of years if you need to build as well the infrastructure beneath so that you are at that level of delivering a data product and this is an investment you have to do before you see the results before you see any value and this is something where there needs to be a minimum of expertise and awareness at the board level to say Okay, we need to invest in foundations so that in two or three years we have those data products scaling for the business to be able to use and once you have managed a data mindset in the
organization data stewardship there is a data governance and you have in each business you have a data governor and a data steward and and people before they type anything in their daily work systems think of data quality once it's established you don't need the co anymore Uh I don't know when is this ever going to happen. Uh right now we need that orchestration. Whether you call it co, CTO, CIO or strategy office, it doesn't matter. Uh but the role needs to be orchestrated from someone in an organization. >> So what is you've seen quite some cool AI use cases and or agentic cases in modern times. What is the current top top use case or what is like the most fun the most business value or the most interesting or tough tough tough agent use cases?
>> So we just roll out or we just the confirmation that we can roll out co-pilot last week um which is amazing because we um needed one year to do all this work also together with the workers council and so on. Um and now the people are really excited to do some work with agents and so on. But first of all you have to um enable the people. Yeah. So yeah now you have to um offer education. Yeah. So that means prompting workshops and also what is the difference between the normal co-pilot license and the let's say premium license where you can uh build agents and so on. And that's why um I would rather make more um um communication about the existing agents.
They are already agents. Yeah. So please start with these agents like the research agent and and many more. Yeah. And to uh get the the feeling what it means to use an agent in your daily work. Yeah. But normally it is always good to start with pain. Yeah. So where is your pain? um when it comes to um organization with meetings or um documentation of your meetings uh or um um risen so traveling costs um you know what I mean? Yeah. Expend travel expenses um these are the pain in your organization and maybe uh it would be smart to start with >> agents in in this case. Yeah. So then the feel the people get the feeling of of using >> um the work together with agents.
>> Okay. >> Yeah. Um but this is this very different to AI use cases on the shop floor. Yeah. So when it comes to the shop floor then it's very different because then we talk about optimization of uh um the forklift for instance. Yeah. >> How about you cancer? Any exciting things? >> I mean as as someone who who uh had the responsibility of the IT department. The first thing I think about when I hear a agents is AI governance and IT governance. Uh because nowadays everybody has power automates and now with shed you can build agents and >> so people just start doing it and here again if you don't have a governness it's going to be a chaos and you are still responsible of having of running the systems of having that liability plus the compliance uh uh piece uh you need to have the overview on what is happening in your organization.
So basically what we had in the well not in the past we still have the the the the asset or service management um uh list now I forgot the name what's service management >> yeah you have to do the same now with the agents >> right you need to know exactly which agents are where um access into which systems you need to have the access rights uh uh really designed uh you need an observability end to end what the agents are doing and the access right should be different from humans and and technical users. So human users and technical users and basically you need to manage both uh uh tenants >> and all of these things it's it sounds I I I hate to say it it sounds like you're again the the the showstopper.
Everybody is motivated. Oh yeah, now we have uh uh now we can use a power automate and everybody is using is building their own agents own agents but if nobody is governing that >> it's um it's a chaos. So tough love segment let's say to stop cancer you know in your book mentioned you talked about understanding humans right even though now your position even further right um from a board level what is one thing a CXO should stop doing today to avoid being an AI tourist uh that's a that's a tough question uh basically because uh CXOs have no time to turn and yeah I mean and we still expect I mean it's it's again what I would what I would suggest now as an adviser which is much easier and reflecting on my my life as a as a CTO it's hell because basically you're the whole day in meetings and then on the weekends you have to to read your board uh uh documents there is no time for learning for experimentation for
reflection for um new things you have have to prioritize it very very specifically in your diary for it to happen. >> At the same time you have to take those decisions. And now answering your question straightforward what what CDXO should stop is um to understand the AI sponsorship as a budget approval mechanism. All right. >> So if you're a 6O, obviously you are the the last approver of the budget, but that's not it. >> All right. So you do need to understand the technology a little a little bit more to be able to challenge as well your team and then and then understand what's what's happening keeping in mind that you won't you will never be the best expert because you just don't have the time.
>> So your team will always be better than you in terms of technological expertise. and then balancing that shift in power. >> Uh um that is also something something new. I mean in the in the past and I I need to say this in the past the you would make the best expert you would promote to a manager >> is that >> that was in the in the past especially in engineering. Yeah. You it's called Peter's principle by the way. You are from the 50s. You're the the best expert. But then you were you were promoted to to manager and those are the managers we have now >> with AI your team is much more qualified >> but you're still the one making the decisions.
So if we don't change the way we understand leadership and and the way we live we experience this leadership it's not going to work and we are seeing it. It's not working because somebody who doesn't understand that technology or even to have the vision how it is impacting the business won't be able to make the right decisions. So we need much more trust in the teams and we need to create the environment for the teams uh uh to build the right stuff. Maybe that points directly to you know Claudia from board perspective what is the one thing CEO should start now >> so I doubt a bit uh your opinion because um you don't need to understand the technology because it's not possible for a CXO member you but you see you should see the potential yeah so you you should get the feeling what is possible and that's why they should start an education with AI.
It could be take three months or two months or I don't know um to go through a journey and to build your own agent and to test it and so on and to get this wow m wow moment wow moment which I already talked about because when they can feel it they see the potential >> and that's enough for for now to give more budget or also to see co-pilot as the new environment what was necessary for everybody. Yeah. So like email or Excel or everything else what is um around your environment. Yeah. In in your work. So um using AI in your day-to-day business is um nothing where you can get out money right now.
Yeah. So um it's it's different to the forklift case because there you can define this baseline. you can say uh it takes this time and so on and you can optimize in this direction but when it comes to the people so to your employees yeah >> yeah then um I think it is an invest in the future of of your education of your people >> so speaking of futures we come buckle up maybe one question we can we can take or two um so 2030 hypothesis right three is the new 10 no 10 is the new three right or something different in this so will the CEO of 2030 30 be a human or just a very expensive rapper for reason model.
What I write here? So Claudia cancer, I want four bold hypothesis on AI. How AI will fundamentally reshape the world by 2030. No safe answers allowed. Okay. >> Okay. >> The first one because u it's really hot. Yeah. Because um let's just look back at the last month. But I would say people um who are on demand. So uh you have a full schedule will get a digital twin so that you can talk to a lot of CEOs with the whole knowledge of the CEO and you can ask him directly. So the digital twin so not the industrial industrial digital twin you have digital twins of people like avatars and so on. And this will be more a commodity I would say in the future.
interesting one. >> I'm totally against that, but we'll discuss we'll discuss later. There's actually even a German solution for it and it was a part of the German in prize. I didn't vote for it because I said this is this is the the end of humanity. >> Uh ethical ethical ethical aspect. Um I didn't >> I say I like it. Yeah. >> Good point. If we're talking about risks and then and then I'll have a positive one. Just just give me a second. Um I think right now we're struggling with the transformation because there is not so much AI adoption. I think by 2030 we'll have adoption. The risk will shift to uh blindly uh uh trusting the systems with nobody challenging them.
So there might be that the human judgment since it's not being trained any day every day anymore because the AI is the agentic agents are taking part part of the of the work that nobody's um is challenging them and and really analyzing that output and everybody's just blindly >> trusting the technical autonomy >> um something positive right okay >> another time something positive so We we have okay >> data and AI governance will be completely automated no automated and will smoothly integrated in all your things you don't need to think again >> data privacy by design and data ethics by design is that data governance by design >> but governance is a human it's a human activity how how would you like to automate it fire everybody know the thing is but governance is something like the text declare so you don't like that so that's Why you take the task of the uh people which is not really favorite and that's why you have to think about these tasks not what is human and
what not what's not so no no no I would say governance will be completely integrated >> let's see let's see >> hypothesis any any the um I would say >> by the way the evidence are just rules no no no No, no, no, no, no. I have a whole different different perspective on governance and I learned to love it after I became a board member with that liability, believe me. Um I think the human to machine interaction >> will um will get seamless. It might disappear even. So if we manage to uh scale agents, you don't need user interfaces anymore. So there is no uh HMI anymore. The agencies is just doing it. So that's would be a shift in our daily work >> and language.
I think uh we underestimate speaking with AI. That's not already common here in >> of Siri and Alexa. Is that coming? >> I would say yes. >> All right. What else? >> We have five already. >> All right. We have five. >> That's true. So maybe we have one and a half minutes left. Any one question out of the audience? Is there something you you eager to ask? Hands up. >> There you go. Am I >> so maybe also >> can I give it for the for the for the stream? >> Hi. So um I don't know if it's a question really but um something like a fear um that I would like you to comment maybe.
So, um, basically I was watching this like movie, it's a little bit older, um, about Sulli, the pilot who landed the, um, plane on the Hudson River. And so, here's the analogy. Um, h, what I didn't know actually about the case is that he got he was like judged in the first place that he was not following the technology basically. And um what I'm afraid is basically if if that might happen to us that we got judged by I don't know a council that we didn't follow the AI rules like in the future. So not only trusting it but getting judged by humans that we didn't follow the AI rules kind of. I I think this might happen really soon in the um healthcare sector.
uh probably not with with transformerbased solutions but with uh with more uh um yeah neural networks in diagnostics for example and we have seen that the results are really good. So it could be that in the future a doctor has to follow or at least ask the AI before making their own judgment and if they don't even ask the AI they might be punished for it. This is uh I think it's the use case where I see it the first happening uh because especially in diagnostics um yeah I don't know do you have another >> I cannot take your fem by the way yeah because the thing is mistakes will happen you cannot expect something from the technology what um human also cannot be or do yeah so we are not perfect and the technology is also not Perfect.
This is the mirror of us. So there will be mistakes. Deal with it. Face it. But we have to learn through the mistakes. All right. So that brings us actually to the closing. Are there may any last sentence you want to give the audience away? uh maybe to to encourage some more uh um uh yeah courage in the decision- making and that is clarity. So in my experience clarity follows the decision it doesn't precede it. So if you expect to have clarity about everything before deciding this is never going to happen especially with AI. So maybe shift that uh sequence. >> Interesting. >> Um maybe coming back to your sentence um we should be more let's throw away all that German mindset and let's be more brave and do something and start something uh and let's say uh take an example of this American mindset which is really cool that's the only one from the Americans we should take care um um the mindset and to start something and be brave that would be necessary
in the field of data and AI Cool. Kenzer Claudia, thanks so much for for spending the time and and you as well. want to give yourself applause um as as you walk out.