Velkommen til Lørn.tech, en lærings dugnad om teknologi og samfunn med Silvija Seres og venner.

Silvija Seres: Hello and welcome to today's edition of Lørn.Tech. I'm Silvija Seres and the topic are artificial intelligence, AI. My guest is Ieva Martinkenaite, was that right? 

Ieva Martinkenaite: That was right. 

Silvija: It's a beautiful name somewhat challenging surname. Please tell us what you do for Telenor to build their AI, where you come from and what you think about AI in Norway? 

Ieva: Thank you Silvija for having me here. Originally, I come from a kind of small country Lithuania and I've been in Norway for about nine years.

I have been in Telenor for three and a half years, and my role in Telenor is first to build, or help build the AI ecosystem in Norway, and help telling the pioneer this process. So, I work in Telenor research, the department that has an AI unit today. We have about 12 people working on machine learning, models and applications. And two years ago, together with Telenor and Sintef we opened the artificial intelligence laboratory, and I was behind that process setting it up and running it. And now we have reached the stage where we invited other big industry partners, the biggest in Norway to join the forces and create a momentum for a country to take the lead in AI.

Silvija: Very cool, because some countries like China have a very clear strategy not just to build and be good in AI but they say they want to lead. Can we say something like that in Norway or do we have to focus on a particular area of application or what do you think? 

Ieva: First of all we have to set ambition. If you look at top industries in the world globally Norway is competing at least in three areas. One is energy, another is fishing and third is maritime. These are the areas that Norway has historically been very good at that. These are the areas that we actually have been digitalizing for years. These are the areas we have talents. And these are the areas where we compete on sustainable solutions globally. So, if you look at what the Chinese are doing, what the Americans are doing what the European countries are doing they actually focusing on AI scalability in areas where they're good at. Let me give you an example of France, in the French national strategy they focus on three areas. One is defense, another is health, and third is smart cities. These are the areas where the French people think they are really good at and they have a good basis to start with. I think Norway has to identify those areas where they would scale in AI, where they will put investments and where they will put strategic with talent acquisition and recruitment, and then where they will be strategic towards data. 

Silvija: I love your angle, because in Norway they had this amazing woman as a prime minister forever Gro Harlem Brundtland, and she where saying “det er typisk norsk å være god”. And they love that of course, and I just find it so constructive because you have to say what you are good that. You and I know that every country in the world including the smallest ones think that they are especially good at something. But you know, exactly what is your strength. And this country has some extreme strengths that you pointed out, and so what you're saying is that AI is really an enabling technology to drive development in those general areas. 

Why and how? 

Ieva: Well, first of all, what AI can do. I mean the AI is technology right. It's the enabling technology that can drive automation and that can drive smart solutions, and that can ultimately drive new revenue streams. Now, you need to be good at some areas. I mean, you need to know your customers well, you need to know your technologies well in order to apply AI in top of it. And then you need to have people in those areas that you can retrain and actually give access to the best software tools that are available. So, AI will not solve all the problems, but they can solve problems for particular industries. And for the problem that they are very well aware of. And I think what we should never forget is that Norway is good at, and I talk from the European perspective now. Norway is good at private-public partnership, and we're not only talking about it, we are actually doing it. If you look how the oil industry has developed from 60s. We had a good business, university and government collaboration. We actually worked it out. And for AI to scale you need to have that relationship up and running. You need to have a policy regime. You need to have an ambition level towards investments. You need to have good talent strategically invest into some talents, and AI is a broad field. If we talk about solving some problems for those industries, there is a specific need for specific talents. So that is where universities and research institutes come in. And then the business, the business brings data and problems. So, if you can combine that all set the direction for competing globally, I think we have a chance. 

Silvija: I think we are really good when we focus and when we get really stubborn about it. And we are just about reaching that point. I was researching in AI some 20 years ago, in the midst of the AI winter. And it's really amazing to see how we've moved past the point where everybody believe this is pure science fiction into the heart of science reality. If we can be nerds for a moment, AI is a very broad field. It's one of the problems when people want to learn AI, do you go and learn the particular machine learning language, because you have to deal with image processing, you have to deal with language processing, you have to deal with psychological modeling, you have to deal with robotics, you have to deal with process automation. So, it's a very diverse field, and at the heart of it there is data and pattern recognition. How would you define AI?

Ieva: Well for me AI is the science of making machines smart. Basically, these are a set of methods that actually do some intelligent tasks, the task that we as humans can do. And the intelligent tasks are what. There is a broad field but at least there are sort of four types. Perception, planning, this is also very intelligent task. Recognizing the environment around you. So, if you look at what machines are learning and if you look what machines can do today, they are very good on perceiving their moment. Image recognition, text analytics. Look at google and what you can do with Google. Google Translate, they are really good on recognizing images, good at recognizing videos. We actually start being a bit better in learning in a very different and very limited environment with games. Actually, machines can learn on the go, they can optimize their actions in interaction with a human, but nothing more than that. Machines cannot plan, and they cannot reason, so that would be very clear based on research today. What machines can and cannot do, so you're absolutely right Silvija, on image recognition on text analytics is just because all of those internet giants have access to texts on social media. They have access to billions of images. They could actually use machine learning to recognize patterns and classify these huge amounts of structured and unstructured data. Other than that, we are far beyond and behind the human intelligence tasks that is just reasoning and planning. So, we have to be always very down-to-earth to understand what machines can and cannot do today.

Silvija: Can you help us with the definitions. I think people should know about narrow AI versus broad AI, because there'll be a lot of discussion related to what they can or can't do.

Ieva: People believe that what artificial intelligent machines can do can overcome human intelligence right. We can be better than humans in all of these intelligent tasks. Narrow AI are machines, models and methods that can solve problems and solve tasks. And I always like to say that when we talk about AI there are two worlds, it is science fiction world, which is general intelligence. And I told you before that we are far beyond that. And there's a real world, the very hard world. The very difficult and research world which takes to put machine learning in operation. This is to me narrow AI, this is actually what solves problems. And I have to say, either you are a big company or a small company or a startup we all have the same jobs to be done. 

Silvija: So Telenor, a very smart giant in Telecom. Moving into digital services more and more convincingly. What are the concrete projects of relevance for Telenor related to AI? 

Ieva: There are three areas where you use AI today. One is network analytics. 

Silvija: Meaning what?

Ieva: We use machine learning to predict our bandwidth. Where are people moving, where they use the internet, what type of network qualities we need to support them. So, we have network data to understand how we can plan the network investments. How can we actually make our network smarter? Meaning more personalized, more agile, more directed to the needs of the customers where they are and where we need our bandwidth mostly. So that is on the network side, I mean cybersecurity comes on top of that. We want our networks to be secure, so we won't use machine learning to identify anomalies or threats that are coming to our network, so this is the network part. We start to believe this is a very unique and very strategic area for us, because that is where we own data. Another era is customer interactions, both when it comes to marketing and sales, but also customer care. You heard about these chatbots, is not something that is new anymore. It's a really scalable application. Chatbots can actually automate our customer interactions, make it faster and smarter. And other than that, the next stage we are involved today is in this preventive care project where we want to identify the customer call and the problem before it appears and fix it. And machines can do this today. And this is still research projects and Pilots, but we're seeing more and more that we can actually put machines and computers to identify problems for customers, and actually act on it. And then the third area is new services and products. We believe that what infrastructure companies or telecommunications can do is to access to new sources of data, and we talk internet of things. We believe that Iot is an area, a new area of revenue potential and new revenue streams for telcos, and we want to use machine learning first of all to get access and capturing that data and classifying the data for different verticals. And we talk B2B markets, solutions for municipalities, solutions for energy companies. 

Silvija: Help us understand what solutions for example municipalities. It’s really interesting because I think what you're talking about is almost the social engineering figuring out what people need from their behavior online and then providing better services? 

Ieva: I can give you example of air quality monitoring case. Air quality is a problem for any municipality, they want to improve it and understand at the level of granularity where the pollution is. What the current sensors can do today is provide you with data on co2 to a particle dust, but not at the level needed. So, we're trying to put new sensors with new infrastructures and new networks where we actually building, such as narrowband Iot, sensors that are powered by battery. And then when we put those sensors in those municipalities so we can actually get the data and try to understand how we can predict the levels of pollution. And how can we actually enable other third parts to make applications, for example for asthma holders. 

Silvija: Or even manage traffic flow?

Ieva: Or manage traffic flow, and information support systems that will come out of that. Out of this data acumen and smartness. To the municipalities or other third parts you can take and make applications. So, at the end of the day we may act as an enabler of infrastructure because we're setting it up, but more than that we can offer some kind of value addit with different support systems for different problems. 

Silvija: We talked about special industries where Norway has an edge, but who are you most impressed by internationally? I noticed that you will be speaking a lot in Asia and what are your perspectives on especially China and Asia strength?

Ieva: There are two things. If I'm asked who I am impressed of mostly there are two countries that I'm very impressed. Canada and Singapore. If you look what Canadians have done for five years ago, first of all they took a bet in AI game and said, we are in the midst of global game and at the top level government level they said we need to establish the top research activities in Canada and lead the game. And if you look what they actually have done, of course thanks to the top three AI scientists in Canada, if you look at the machine learning deep learning scientists, they are top there. But what they have done, they actually created labs and ecosystems quite successfully. So, my advice would be to the Norwegian government go and visit. And I know that there are ministers that are visiting that echo system. The second thing is Singapore, it's a different model. It's a top-down model. It's a government driven model, where they also realized “how can we compete against China in five years?”, and they said, the only way we can do is to be smarter. We don't have natural resources so the only way we can compete is being smart. And artificial intelligence came naturally to them as kind of the enabling technology for three industries, and they said in three industries we'll invest. The first is Health. the second is as smart cities, and then the third is finance. And if you look at the speed and at the level of investment that they are putting its amazing. 

Now China, they have created an AI first strategy. They said we will lead. There's a three-year plan and we will lead. Now there are leading in several areas. But what they are investing in is industrial machinery and production, plants. And that is where we probably very hardly will compete. The second thing, if you look at what is actually becoming worrisome for the global world. If you look at the top two conferences this year in AI and machine learning, one nips in the US, and Arch guy in Sweden. If you look at out of 500% of applications, 70% were Chinese. So, they're actually moving and cutting the research. So, its not only applications that they're good at, their starting being very good at R&D research. So, they really bridging the front.

Silvija: And connecting this to genetics which is where things get super interesting. Eva, I would actually love to speak with you tons about AI and ethics. We are running out of our time so I think we will go towards the end. If you would like to leave people with one mental image, this is a podcast and so they need some images in their heads when we are done talking. What would you like to live with them related to AI opportunities? 

Ieva: I would probably have an image of and AI podcast saying that AI is going to change the world, and then there are two leaders who are saying “No, I'm not concerned about it. I don't think it's coming”. 

Silvija: You have heard this obviously. 

Ieva: Yes, I do, and I'm concerned about this because it's already now. It's in operation now, but I think our readiness and I think in General our readiness and educational level of understand what AI can do and can't and how that can affect us is still very low. 

Silvija: We should get very busy. As they say future is already here It's just not evenly distributed. And I think your examples about Canada and Singapore and some of the concrete industries, what others are good at are incredibly useful, because they focus people and understanding. You can't move the ball on all fields at once, but there are some areas you can really move it.

Ieva, thank you for enabling this AI work both in Telenor and in Norway, and thank you for teaching us about AI in general. 

Ieva: Thank you so much for having me. 

Silvija: Thank you for listening. 

Du har lyttet til en podcast fra Lørn.tech, en lærings dugnad om teknologi og samfunn. Følg oss i sosiale medier og på våre nettsider Lørn.tech.