Navigating Responsible AI: A Conversation with Thomas Nørmark

Speaker 1: Welcome to IBM Partners in AI Innovation. This is a unique podcast that focuses on two areas of how artificial intelligence has been put into practice. First of all, how ecosystem partners across Europe, Middle East, and Africa are playing a role in innovating around AI and then secondly, how they're doing so with trust and responsibility. In today's episode, I have the pleasure of discussing with Thomas Noermack, who is the Global Head of Innovation at NTT DATA about the role of the inaudible system in helping businesses achieve that real competitive advantage, the real business benefits of artificial intelligence. So Thomas, welcome to the podcast.

Thomas Noermack: Thank you very much. Happy to be here.

Speaker 1: Okay, so thanks for joining me. Look, before we delve into the discussion today, I think it'd be good for our listeners to learn a bit more about you and the partnership of NTT DATA and IBM, which we've made an important announcement about today.

Thomas Noermack: Sure. So personally, I'm the Global Head of Innovation in NTT DATA business solution, meaning that I'm overseeing all our local country organizations and trying to build up the innovation practice in different countries. NTT DATA is one of the biggest system integrators in the world, and we are more than 200, 000 employees in more than 50 countries. So we are really trying to leverage on that global strength, but also to create the synergies and make sure that we don't invest in things in parallel. We only want to pay for things once, and then we want to distribute what we have built to the other countries.

Speaker 1: So let's just go right into AI, which we know last year there was a huge buzz around AI and generative AI, obviously in particular, and many businesses were looking at, look, how do I convert all of this buzz, all of this hype into some sort of business reality. But actually what we're hearing from a lot of clients, a lot of our partners, and I know you've heard it too, which is actually what matters here is responsible AI. And I wanted to get your perspective on why that should be, if not one of the first considerations for any technology decision.

Thomas Noermack: There are various angles to this. One is of course the legislation and the laws, which we are seeing, especially in the European Union now with the AI Act. So companies are actually required to comply with different principles within responsible AI, ethical AI, also sustainable AI. We have been working with this topic for many years, maybe it started a little bit as a coincidence because one of our first real AI/ machine learning projects were for a helpline in Denmark, inaudible, and the helpline was actually for children. So children could call into this helpline and then they could get help with the inaudible ride. And when you are dealing with AI and children at the same time, then you have to really think about these principles.

Speaker 1: Now, the Center of Excellence, big announcement today that those who have and seen it's a Center of Excellence around Watsonx, our AI platform for business. Can you tell us why you did it and what will this deliver for clients?

Thomas Noermack: Sure. So actually the timing was perfect for us because we were working with the general purpose large language models like GPT-3, 3. 5, GPT- 4, and so on. And we could see that this was really not enterprise- grade AI for us because of hallucinations and copyright issues, privacy issues, there are a lot of pitfalls working with these models. So when we heard the announcement from IBM when they announced Watsonx. ai and the concept around training curated domain- specific language models where you would not have problems with hallucinations, you would not run into copyright issues, lawsuits and so on, then we were clapping our hands because this was exactly what we needed at that point. We had a lot of different and interesting opportunities, especially in the banking sector and financial services. And we really wanted to look into using these domain- specific last language models that have, for instance, been fine- tuned and optimized to know everything about financial services and all the flavors and nuances of financial services. And this is what we then got from the Watsonx. ai. And this is where we immediately reached out and says, " We want to collaborate with you IBM on this journey, and we want to be the first." And actually we then agreed to set up this Watsonx Center of Excellence where we would enable a lot of employees, but we will also bring in a lot of customers. And then together with IBM, we will actually start implementing real- world projects using Watsonx. ai. And this is what we've been doing for this past six months.

Speaker 1: Yeah, no, it's been a fantastic collaboration. And I just wonder, you say six months now, are there any use cases you can see emerging that might be prevalent?

Thomas Noermack: Yeah, so the first, we started with two projects and now we are then wrapping up with four other customers. And we could see that for the first projects, especially the combination of using the open source AI models from the Hugging Face library and combining it with the Watson models and the governance part of it, that was really powerful. And this is, of course, we are still working with the GSI lab and with the client engineering of IBM to fully understand how we will combine this in the best way. And of course also as being one of the first, we are also providing feedback loops together back to the development teams to make the offering of the products even better in the future. Yeah.

Speaker 1: You mentioned the GSI labs and client engineering teams. For those who aren't familiar with that, GSI labs are our system integration labs. Many are based out of India, but all over the world who work with our integration partners to be able to build solutions. And then our clean engineering teams work on the ground in most of our countries across Europe and in fact around the world, of course, which are working on practical use cases that are designed for specific clients. So great to see you taking advantage of that talent. I wanted just maybe to pick up on a similar point here around the whole area of risk possibly, and I wanted just to get some thoughts from you in terms of advice you give to clients around the threat landscape, particularly where you start to use data in GenAI applications. That data can be stored in different locations coming from different sources, obviously from third parties. How do you try to make sense of all that and minimize the risk of, let's say, particularly when it comes to the proprietary nature of some of that data, personal data security, and then the potential ongoing risks that there are that are unbiased and prejudice in decision making?

Thomas Noermack: So in the first projects we develop around this, we build all these mechanisms ourselves with explainability, monitoring the models, making sure that they're not drifting, but also the business monitoring, making sure that the accuracy or the performance of the models still be strong. And then we learned that this was a huge task to do this in every project from the bottom up. So we then partnered with IBM around their Watson governance tool, and we have done it in the past with other providers. The problem was that it was on the terms of the governance tool. So we had to migrate our data and our code and everything onto this platform, this governance platform in order to get the benefits out of it. But what we can see with Watson governance is that it's not intrusive in the same way. So you can actually use it more like a sidecar or kind of plug- in that will not interfere with your existing code.

Speaker 1: I'd like to just finish with, and by the way, you'll hear a lot of noise here because actually I need to say this. Thomas and I are here at the inaudible conference in Barcelona. So a lot of buzz here where you did an announcement earlier today, right?

Thomas Noermack: Seriously.

Speaker 1: It's fantastic to see some of the technical community coming together here, and there is truly a lot of buzz around, particularly around Watsonx as Thomas and I talk here, but maybe just opening up more broadly to finish with, what do you think? I mean last year, the incredible amount of buzz, I mean look, AI is not new. The AI technologies and capabilities have been around for a while. What is new is the concept that I hear of foundational models, right? Large language models. What do you think has to happen? The one thing, and I'm going to really pressure on one thing that has to happen in order to take the concept, the technology that that's there today and really turn that into pervasive business value, let's say, to get us to the promised land, if you will, around artificial intelligence of business.

Thomas Noermack: So I would say, yeah, it is a big question to give a short answer to, but yes. But right now we see that the technology superpower has arrived, right? So it's not about the technology anymore, it's really about the change management that corporations need to implement, both when it comes to technology adoption on a personal level. So I'm using my GenAI agent, my GenAI tool to improve my personal efficiency, but also when it comes to actually using the technology in the offerings and the products and the services that you having in the market. The change management part of actually securing the adoption and the willingness and the understanding of the superpower of these technologies, I think that is something that companies would have to look into over the next period of time.

Speaker 1: Thomas, look, it's been a real pleasure to have you on this first edition of IBM Partners in AI Innovation and in particular, the leadership bill showing and NTT DATA is showing here in Europe in bringing AI to the market responsibly. And it's fantastic to be partnering with you, and I just want to thank you for your time.

Thomas Noermack: Thank you very much. It was a pleasure.