Christian Guttmann is Global VP of Artificial Intelligence at Tieto, an IT software and service company providing IT and product engineering services, with approximately 15,000 employees, active in around 20 countries.
Christian talks about AI and wanting to understand intelligence enough to recreate it.
Christian discusses building systems that can interact with humans beyond regular computer interfaces.
He started working with computers early, too early for advanced AI. He studied broadly, including philosophy, ethics as well as synthetic and biological neural networks.
Christian has found a lot of inspiration in the old papers by Alan Turing and others. He sometimes envies them their opportunity to think big, as they did when they founded the areas of computer science and artificial intelligence.
Christian has be focusing on AI in healthcare and has recently started to communicate the opportunities and challenges in artificial intelligence to the general public. This is something that the host Per Sjöborg is also very passionate about.
We also get to hear about the Nordic AI institute (https://www.nordicaiinstitute.com) and the work it does to inform all parts of society about AI. Anyone interested in AI is welcome to reach out if they have questions or if they have knowledge to share.
Then we hear how Christian is working with Tieto (https://www.tieto.com) on integrating artificial intelligence in their customers daily operations.
Per Sjöborg, host of the Robots In Depth podcast, interviews Christian about AI. Below is a transcript of the interview.
Christian: I think it is an enormously fascinating concept. When I was younger, I started my journey into artificial intelligence and robotics. I think I had two questions on my mind. One was on one hand to understand how intelligent systems work and that brought me to studying psychology and social science then of course also engineering. If you want to understand intelligent behaviour you really only have understood it if you can engineer it. I also did those types of studies but I think I was fascinated by the fact to build something with an intelligent capacity, something you can interact with, something that can interact with the world or with me in a very advanced way. That fascination has carried through from an early age until today I would say.
“If you want to understand intelligent behaviour you really only have understood it if you can engineer it”
Per: Can you tell me how you started? Did you study at university or how did you get into the field?
Christian: I think it was early on I had already a computer when I was like I think 10 or 11. I started doing my programming at that time. I mean with these small computers you had in the old days there is no way you can fulfil the big ambition.
Per: No deep learning on those.
Christian: Not quite although many of the theories were already in place but then I knew when I started studying that’s when it became much more formulated. That brought me really number one in the field of theoretical computer science, also philosophy and ethics early on. That was a good while ago.
Per: Then you discovered AI specifically and started to focus on that.
Christian: Exactly. I think my first course in AI was in 1996 or 1998. At that time the plethora of topics that you have in AI it’s just overwhelming, so fascinating about how you can apply these types of technology including the good old networks of neural networks. When I studied these topics it wasn’t just like artificial neural networks as we know them today but also biological neural networks. I was actually in courses where the neural networks, we were looking at them at a deeper level really hold a function in our brain or in invertebrates for example so I was very fascinated.
Per: Is there anything that from those early steps into biological brains that you’re still thinking wow, it’s still relevant?
Christian: Of course, many and many of the basic questions are also not answered yet. I mean we still have a lot of insights. Recent advances in neural networks also point at these early structures how you can for example over time maintain knowledge in neural networks that otherwise would be forgotten. That is very similar functionalities that our human brains have. I think much of the research today has very much influence of the early days. Again, even though I started in 1996, 1998 as you know AI, robotics, those areas have started much earlier in the forties and fifties and folks like Alan Turing have been really opening up these discussions which have influenced me a lot too. When you look at the early papers by Turing they are probably some of the most inspirational for me and then of course not only neural networks but robotics and all these areas. When you start seeing the challenges in each of these areas it gives you a lot of respect and I think when you then also connect this to social science and to psychology how we act it becomes enormously fascinating. Are we ever able to build such systems that have these types of abilities or some of these abilities? I like challenging problems, complex, challenging problems. If they exist I will be there.
Per: Yes, big problem we are talking about and AI certainly is. That is a slice of the problem pie I can tell you.
Christian: One thing I must say, when I look back at the old papers from the fifties, sixties, seventies, I’m a little jealous to see how at that time there was much more freedom and liberty to propose much bigger goals and ideas which today if you would have the same propositions they would already be discussed. In other words at that time the goals and the visions projected were much bigger whereas today, also when you go to conferences like here at the International Joint Conference of AI problems become much more tight. It’s a lot about mathematical formulas. Sometimes it’s also connected to really bringing lots of papers and publications forward. I think it is important sometimes to reflect back on what the big goals are that at least motivated me and many in the field that have dedicated to it and have been working for it for decades and will continue to dedicate their life to it.
Per: What you’re saying is absolutely true and I think we must take time in our busy lives to do that, to raise our eyes above the details which are of course in the end very important but we also have to keep an eye on the direction. Where do you see both your work in AI in general, what is the direction? What are the problems that are attractive for us within a reasonable timeframe?
Christian: I was just reflecting about what you just said earlier. You are right. When we look at what our next big achievements are having a certain discipline about what is needed to achieve that. When we look at science, it’s a scientific principle that enabled us today to really discover more truth about our world and that is the Age of Enlightenment. It’s not necessarily that one generation is able to come up with the way of really fundamentally changing the structure of society if you like. It took hundreds of years for us to develop these mechanisms and similar in AI, we need these skills, these understanding. Right now it’s a very high level of understanding of what is happening. I really hope in the next 10 to 20 years there would be more.
To your question, what is it that I will be doing. Number one, there is of course more and more research that needs to be done. That is at the universities that I’m associated with. With students we had a really fantastic paper on how to apply AI in the healthcare context.
Per: Because you focus a lot on the healthcare context.
Christian: Exactly right so I think over the last 15 years I’ve been focusing mostly on medicine and health and that is a very fascinating topic but at the same time now I feel a huge need in making a contribution on the clarity of the topic of AI so that for example governments, large corporations and also start-ups and the whole of society understand the topics better. It’s a big dedication and it’s important why we talk because I think of course media and communication is extremely important in order to get a good picture out. As is often the case you need to have these clear views about what the topic is by perhaps people that have been working on it. I’m a little bit surprised in other areas you would pretend to have experts on the topic of let’s say brain surgery or some topic and have everyone speak about the topic without having the qualification to speak about it. I think in AI it becomes even more important because of the decisions that machine learning algorithms will do. By the way your fantastic discussions on the topic will be fundamental. It will influence our lives and as you say, if you run it to the wrong direction and governments or companies place their bets on the wrong horse so to say, it’s going to be really bad.
Per: That is very much why I run the show. This aligns perfectly with what I’m doing which is in part why I’m so very happy to do it because I carry the cameras and these guys know their stuff. It’s all about you. I give you this venue to share the information you have and you know that this happens without any filter. You know that. That can get correct information out there. It can also dispel myths and make this process better by informing people.
Christian: This is exactly what I try to achieve with my mission and also the Nordic Artificial Intelligence Institute…
Per: Introduce the institute. What is it? Who works there? What do you do?
Christian: Briefly it’s a non-profit organization. An independent organization which is a global alliance of leaders in AI that both have a strong business representation and leadership so often global leads and global partners and large organizations around the world. It’s international and also those that have a strong scientific affiliation so often are professors or PhDs and or have an association with governments. They are for example in the UN or discuss with governments what is happening, those backgrounds that alliance of leaders is in the Nordic Institute. We are essentially offering independent advice on anyone that needs it. At this point it’s about clarity and the purpose and the mission of the organization is to use AI for social and economic prosperity in the long run for us as humans.
Per: A trusted third party for information basically. There you can reach people. We can encourage the audience to reach out and they can find people with deep scientific knowledge in many areas. I guess also people can involve themselves in this and share their knowledge and experience and expertise.
Christian: People are welcome to become members of the Nordic AI Institute. Then we have NAII fellows. They are now very active and also giving interviews and proposing and contributing where we can right now. There is a lot of demand. A lot of countries and companies, for example are also building AI strategies. That is something where there is a lot of questions still around and mixing up for example, let’s say old style business intelligence or the things that have existed as opposed to the actual AI machine learning algorithms that are now actually, fundamentally, differently influencing business.
Per: Let’s go back to you. Can you talk a bit about your daily research? What are you spending the days in?
Christian: That is a good question. In fact I have many tasks. I am the Vice President at Tieto and Chief AI Scientist there. That means that we are actually building up our capabilities of delivering artificial intelligent systems. Within products we are actually building verticals. My unit at Tieto is very much around innovation. We are looking at partners on building products over a much longer lifespan if you like. We focus a lot on machine learning for example, deep learning of course. It’s big and I think this is something that we have only started our journey in the Nordic countries and globally at Tieto where are using the data in order to really improve a product.
That is one and then we have machine vision and natural language processing products. There we have a top team. Most of them have a PhD in machine learning or AI and I’m super privileged to work with such a leading team here in the Nordic. I would argue that we are probably one of the leading AI teams here in the Nordics in terms of really producing tangible results in this sector. That has also been broad. I know of course many of our friends in this sector are working on similar things, sometimes a bit more specific on telecom or in cars and so on but we have a very broad perspective.
I should mention that we have smarter society as a concept that we are driving at Tieto overall. All 16,000 employees are very much focused on delivering smarter society as a concept which means that we are more able to collaborate with each other that many of the tedious tasks that we are doing are sort of easier.
Per: Also allocate more resources more efficiently and also a society as a whole is never going to work optimally. It’s always about taking this traffic accident into account or this reconstruction and then trying to dynamically because it changes all the time. If you have a static plan you’re doomed to failure. You have to constantly look at what is our goal and how can we best reach it given all these things that suddenly it’s like 40 degrees in the shade or there’s a snowstorm or critical infrastructure beaks down for some reason and we still have to handle it.
Christian: Now you’re actually relating to very interesting research topics that I probably work with for a good 10 years or so with Prof. Michael Georgeff, who is also the director of the Australian AI Institute. We use the same technology used for space shuttle BDI architecture. These are exactly these hybrid planning approaches which are then also used in chronic disease management because you cannot have a static plan for example and then expect that you execute. Michael actually often used this example of travelling to the airport and then you have a flat tire. If your plan is to just sit in the car and drive to the airport you will just be stuck. You need to start having plans that you need to execute. That is actually the essence of these types of approaches where you are insuring you’re flexible to the future.
Per: You work with the Karolinska Institute too I understand.
Per: Could you talk a little bit more about what you do together with them?
Christian: At the Karolinska Institute and also at the University of New South Wales in Australia we have topics in research related to how we can improve health, how we can essentially deliver better medical care. That is a primary goal but now with some of my students and colleagues we’re working on neural networks and deep learning in order to predict complications with patients that have multiple chronical conditions so rather than having just one chronic condition often what happens is you have for example diabetes and problems with your liver and so on. If you start taking all these together it becomes significantly more complex situation for doctors to deal with and clinicians to deal with. For example a (0:14:56) is a unit that takes together those different conditions.
One of my students Rafiq Muhammad, he has been working on this field and what it means is that it will be much better for patients ultimately. They will be treated better and more effectively and it will make the life of the clinicians more easier but essentially knowing what would potentially happen, how clinical processes and clinical management can improve in those types of situations. The bottom line of these types of projects, it improves lives.
Per: It saves lives.
Christian: Exactly yes and that is what I’m ambitious and passionate about but I think with these technologies we can make a huge difference saving lives and improving the existing lives and just make the entire humanity get us to the next level.
Per: It’s the next step for humanity. I understand that these multiple conditions and multiple medications is very hard because we get unique so quickly. Say that you have one disease. Sadly enough, quite a number of people will have that disease. Then you add a second one and then the group that has both of those two and then for other reasons might be taking medication against those. When you add the medication in two diseases suddenly that is a very small group. When you add a third condition and medication for that you might be the only one ever in the world that has had that exact combination of medications and conditions and circumstances and then it’s very hard for classical techniques to work because there is simply so little but deep learning and an AI system help I understand.
Christian: Exactly. We are a little bit away from really implementing it and applying it today but you are completely right. It goes to the concepts of precision medicine also.
Per: Individualized medicine.
Christian: Exactly and the reality is that in the next, I mean already today, there’s so massive amounts of data that you should be reviewing in order to make a more proper treatment plan for example for patients and that includes your genomic background, microbiome data perhaps, your entire medical history and also that of others that are similar to you. Once you start having all these information range, it’s understood to be impossible for an individual human being to go through all this and then come up with an accurate prediction. I think in five years’ time we will be seeing more and more of these types of applications in the healthcare system. In 20 years’ time we would be seeing it quite regularly applied everywhere. In 40 years’ time our children and grandchildren will look back and say how could they have done it back in those days.
Per: I really hope that 40 to 60 years now, the kids then are going to look at this period as how could they ever survive without an AI that helped me but if you can help focus the doctor’s energy by the AI sorting away all the things that it’s just certain that is not relevant this time we can focus the human effort.
Christian: I think there are certain aspects that won’t go away. The warm hug that the doctor or the nurse would give in those situations, these are important aspects that will be there for a while. I think everyone has the ambition to do very accurate diagnosis to be effective with clinical management to save money for tasks that no one enjoys. You see so many clinicians making notes and dealing with coding system. That is arguably really not something that contributes directly to care for example so it’s very important.
Per: I heard a really good thing where they turn the AI concept around and say augmented intelligence rather than artificial intelligence. Artificial intelligence augments that human intelligence. We talked about how we in the future will need the skills to use the tool AI. Do you see that we are getting there where we start to train people to use these AI’s?
Christian: To some extent we are already there. I mean the concept and technologies and AI have been out and this technology is already out there when you start using advance systems in some sense. People are used to it. Then of course you’re very right. In fact one of our fellows and Professor Milind Tambe, he wrote a really fantastic paper in 1996 on Flexible Teamwork. It’s all about how you have autonomous systems working with each other and the work that followed after that is how do you have human actors in teams and human robotic teams working with each other. I think it is very much the case to look into this. What decisions are difficult enough to suggest that you have an AI in the system too for those stuff?
Then of course when you say augmented, what comes to me is it depends on who is making the decision here and why the decision is made. When you have today people that make decisions that influence our lives and how it impacts our lives they are the more tricky ones. We need to be very careful. That is the situation when AI might be limited in how it could actually be applied in those types of situations but at the same time augmented artificial intelligence is a possibility for us to really look at the bottlenecks in the system and where we can really expand this knowledge. Healthcare and medicine is one of those beautiful examples where you can use the AI machine learning to understand huge amounts of data quickly.
Per: Do you work on any other projects other than the medical ones directly or together with Tieto? Tieto has a very broad portfolio.
Christian: There are many of course. We have in the banking. We start looking at the areas around loan scoring and repayment cycles for example. There is already a lot of AI machine learning applied to some extent. This is now something that we continue to work on. Retail is another big area. Many of us have seen examples from the US, from large American corporations really looking at automating even more aspects of the retail space. When you go into a shop and it automatically detects what you’re picking up and you don’t have to pay anything with your credit card. You have those types of things that we are working on essentially, inventory detection. Forestry is another big field.
Per: Very big in Sweden, forestry. I would like to hear more about what you do there.
Christian: Very big in the Nordic world. It’s a lot of world.
Per: 40% of Swedish GDP I understand.
Christian: Paper and pulp mills they are important aspects and it’s all about understanding what the trees, health of the trees for example, where these trees require thinning and those types of questions you can start answering or reviewing by having satellite images or other sources of information or drones, of course, is an increasing topic. Also in Australia, by the way, I’m Australian too. I was in Canberra recently, our capital. There you see on farmer’s markets, I was surprised, I was expecting tomatoes and chicken being sold but there were three stands that sold drones. These drones were self-flying and they would enable checking the fences and checking if the sheep and the cows are in the right order. It’s close to completely automated. That was a year or two ago. Those types of similar technologies are used for forestry of course. It’s all happening.
Traffic flow management, when you look at objection identification which by of itself is a large field. You can observe many different objects but looking at how cars and traffic is evolving by looking at how cars are moving.
Per: To predict queues and to manage red lights and other control devices. It’s been a fantastic discussion. We have learned so much. I’m sure we’re going to come back now and then and see what you’ve been doing since in that time. Thank you very much for taking the time to do the interview.
Christian: Thank you Per. It was great. Thank you.
Per: I hope you liked this episode of the podcast version of Robots in Depth. This episode is produced together with Wevolver. Wevolver is a platform and community providing engineers informative content to help them innovate. It is how engineers stay cutting edge. Aptomica is the founding sponsor for Robots in Depth. Aptomica runs anything in modular robotics. Dream, rent, build. Visit Aptomica.com to connect. I am your host Per Sjöbor. Until the next episode thank you for listening.
END OF TRANSCRIPT
The Wevolver Robots in Depth podcast is published multiple times per week. If you, or your company is interested in supporting this podcast and reach an audience of professional engineers and engineering students, please contact us at richard [at] wevolver.com