 I'm here with Ilka Tumi, who is going to be our keynote speaker on day three of the Eden 2019 conference here in Bruges. Ilka, I'll be fascinating to hear what your keynote speech themes will be about. Yeah, it's going to be interesting, because I'm trying to cover both the big picture and a bit more about the detail. So I'm going to talk about artificial intelligence and the transformation of education, looking how the 20th century educational systems are being changed because of wider use of technology, digital technologies, and specifically artificial intelligence. So I'm looking at the kind of future of work and transformation of social and economic structures. And in that context, I'm trying to see how education can respond to those demands that will be important in the future. So that's the kind of big picture. So I'm also saying that the devil is in the big picture. On the other hand, I'm talking about the detail, because you also need to understand what these things that we call artificial intelligence, what they really do. There's a lot of newspaper articles. There's a lot of kind of hype around this. There are huge policy pressures to do something about AI. European Commission is now planning to gear up the annual investment in artificial intelligence to 20 billion euros. And they are investing a lot of money. All different continents are doing the same thing today. And everyone is saying that this is the biggest thing since the invention of electricity or oil. Or I even heard UNESCO director saying that this is the biggest change in human history since the Paleolithic time. So there's certainly huge expectations, kind of policy level expectations. And we, of course, have all read and seen Hollywood movies. We have read science fiction and seen Hollywood movies. So we kind of think that we know what AI is. And we project these imagined AIs to the future and think that everything is going to be different. It's very important to understand what these systems actually do. So they are very limited systems. They can do very interesting things, but only very few interesting things. So in a way there is a kind of cumbrian explosion currently going on that some very simple basic principles are being applied in very many different areas. And you get really exciting kind of outcomes. You can see Mona Lisa speaking and talking and you can do fake news and you can do all kinds of simulated humans reading news reports. You have robots that are able to jump and dance and you have all these kinds of exciting things and cars and autonomous vehicles that can drive on their own. So it seems that there is a lot of happening, but the underlying technologies are really very few and only kind of specialized. So what I'm going to also talk tomorrow is about the relation of learning theories to this AI. So artificial intelligence in name is about intelligence. So for learning theorists and educators it's actually quite interesting that what these guys mean by intelligence. On the other hand the current hype in artificial intelligence is about what people call machine learning. And again the word learning is there. So it's kind of interesting to see what kind of learning happens in these systems. You talk about autonomous vehicles, you talk about Mona Lisa talking. These are things that are very entertaining and very good in terms of getting us around the world and so on. But what about education? What aspects of AI can we apply to education? Well there are very many applications already currently and obviously because artificial intelligence is high on all agendas now. Many digital technologies are sold currently as artificial intelligence. I mean it's difficult to sell anything to schools in the next couple of years without claiming that it has some kind of artificial intelligence. The problem of course has always been that for teaching technology never is an answer. So it's a combination of teaching practices with kind of material tools and things that make teaching possible. So whether it's blackboard or whether it's a textbook. And what we are currently doing with AI tend to be solutions to the kind of current problems in teaching. So for instance one area which is very popular is automatic assessment of exams and tests. Because currently we do teaching for tests quite a lot. So in many countries it is really very important socially to succeed in tests. High stakes testing is a big thing. But it's only in some countries. So it also should be understood that the ways that we use these technologies are very different in different cultures and different contexts. There's a certain Ray Kurzweil who talks about the singularity event when we get to the point where supercomputers become more powerful than we are. Do you think we'll ever reach that point in our history? Well we have supercomputers where they called that. The answer to that question is actually quite illustrative I would say. Because if you look the learning models that current AI systems this kind of data driven or deep learning models use. They are from learning theory point of view they are what we call associative learning models. These are something that all living beings actually use. So these are low level learning processes that are possible for insects and pigeons and rats and so forth. And that's the only area where the current AI systems this data driven AI systems work or operate. So they should be called artificial instincts and the learning models are really similar to what biological kind of beings develop during millions of years or thousands of years. So they are kind of reflexes and the learning models are purely behavioristic. They are like thorn like or skinner type of models. They are really that's the level where they operate and they cannot move beyond that. There is another kind of artificial intelligence which is very often used in education. I mean for the last 30 years most of AI in education has been on what people call symbolic AI or expert system based approaches where you represent knowledge and then you match these basically so that you can do let's say intelligent tutoring systems or something like that. So the symbolic AI is a different animal and it's not on the newspapers today. The thing that you see today is data driven AI which is this kind of very primitive instinctive learning model. Symbolic AI is a bit more complicated. That's something that theorists like Piaget maybe was talking about. Conceptual development, conceptual models, cognitive science is very much for focusing on that during the last decades. What we don't see today is the higher level cultural learning and that's the area where like Vygotsky or Paolo Freire maybe would be talking about how culture makes us, how we learn from the more culturally kind of competent adults, languages, conceptual systems, sciences, cultural behaviors, social practices. That is a level where currently AI simply does not exist. And the Freirean question about how we actually change the society through learning, that's obviously something that is nowhere visible today. Do you think we'll ever get there? Is this something that we will ever achieve with machine learning? Well, if you say that an insect can learn something quite complicated behaviors using associative approaches, you can climb up to some extent. But learning theorists like Vygotsky from the 1920s already emphasized the point that there is a qualitative change when we become competent humans. He was kind of comparing like learning in apes or primitive cultures and non-primitive cultures in the 1920s, which meant cultures where you have conceptual systems like sciences and so forth. So these are cultural artifacts in a way, cultural systems of meaning and we learn them in social processes and that's something that you cannot do instinctively. So that is an area where the current kind of technical architectures of AI in my view cannot achieve. So that's a level of meaning. We can do kind of behavioristic reflex type of responses to changes in the environment. That's what the current systems can do to some extent. But to create those environments and to operate in meaningful worlds, that's something that current systems cannot do. That's the area of learning, of human learning. Okay, it's been fascinating talking to you about AI in the future and education of course. Thank you very much.