hey great I did so I start off by saying straight away they did a pretty dangerous thing which is yesterday I changed my presentation because Fabian said hey make it a bit more TED talk like so so it's pretty new set of slides and I see if I can can pull down a nice storyline from those slides and I think I would start just saying a few things about myself so I have been working in the field of AI for very long time 25 years started off doing my masters and PhD in the field of AI and it actually I started off doing my psychology degree because I found it terribly helpful to understand human behavior and animal behavior so how do animals and humans make decisions under which type of context and took it all the way through becoming an adjunct associate professor in Australia but at the same time another big thing that I did is of course having my career and big companies in the world so I was at IBM for
example at the British Telecom you look Packard large big organization now I'm the vice president global head of AIT at OU so I've been doing hundreds of these types of AI projects very passionate about it and then finally I'm also an entrepreneur and I guess there for example with Fabian and I have a bit good bit of overlap here and also VC funds and so on looking at the next big thing in the in the entrepreneurial and startup world which I find terribly fascinating because usually all the innovation and this risk-taking happens in that sort of space so so that's a little bit my background and I'm course extremely excited that AI has become such a big topic in the world looking back you know it has been quite a nerdy topic so I mean some of you can probably confirm that it's Pro in the last few years has become a forum for those that were the absolute Nerds back in the days right and in fact now they're switched from being a topic
which only a few people talked about and knew about to almost everyone talking about it also those that are not necessarily qualified in fact there are many that use it now a little bit not only a little bit but a fair bit as a marketing type of hype right so startup certainly attract a lot more funding and companies saying they know what to do although I'm speaking to a lot of CEOs and and also ministers and it becomes very apparent that that knowledge is not truly penetrated in the in the mind of those people which i think is a bit sad but okay just to think what has happened so far so just asking the people in the audience to get a sense of what has happened in the past and what children today Millennials have no experience with so how many of you have listened to music tapes like little tapes that you put on tape recorder and so on so very many right so that sort of technology is obsolete and you see
probably on YouTube a fair few people have fair few kids you give them that tape and they have absolutely no clue what to do with it right how many of you have used a fixed telephone line to actually call up people and arrange things in your lives very many what does that say about the age average here in the room you know but look yes again you yeah yeah but you know very again you find very entertaining YouTube videos and you get 18 year old kids you know all these old telephones there absolutely no clue how to how to use it yeah and so that's just giving a bit of a sense and those types of technologies are completely foreign to the kids in the generation today and when you look into the future you know what type of things are we used to today which kids won't be in the future our kids won't be exposed to and one example is so how many of you have ever actually kids younger than five years
old so about twenty thirty forty percent whoops destroy it so so those kids are most likely they won't be learning how to drive a car they don't need to so in 10 15 years time there's very very highly likely going to be cars that will be self-driving you know those types of technologies are never meant to be on the road like next year the year after they take a little bit of while to come through so they will also be living these kids will be living in a hyper personalized world so everything they will be receiving will be very tailored to them so it will be for example articles which are not generically written and for example that cited site or in in other newspapers in Germany on in New York Times but it's very likely that not only the channel at which you pre get presented this type of information but also the content itself is actually going to be highly individualized to your needs and who you are not necessarily the message the
content of the message but how it's written another example is you know precision medicine so what I'm seeing now and I'm also at the Karolinska Institute which is one of the big medical universities and Stockholm which gives out the Nobel Prize and in medicine and you know what I'm seeing now is that they are already AI technology use and they in the healthcare fields for example some of your way of MRIs and and those types of pilots as MRIs and x-rays where you start seeing abnormalities and the scans and my prediction is that in about five years we see some of those technologies being in place and twenty years it will become in place and in 40 years our kids will look back and our grandkids will be looking back to the state today and think how did these people do that so how could the GP possibly make sense of this enormous data that you're representing as a person you know so that that's just when we look back to 200 years ago and
how you know in the middle age medicine was delivered right kids and our grandkids in 40 years we'll look back to our time today and see well how could a GP have possibly made sense of all this genetic data all that information that the individuals have in terms of a historic medical history and how could they have compared that to other lots of other types of patients who have sort of the same type of background right very complex and they are in fact quite a few examples and I will mention that in a minute of people that have similar conviction so and and the other thing is and I think that's sort of the main message of the talking some sense they will also be interacting much more with with a ice so there there will be these human mixed AI teams and the technology that you're using and the way you interact with it will be much more much more important so there will be chat BOTS and mutual assistance there's already plenty of
cases in China where people are starting to form a very close bond to some of these chat BOTS and virtual assistant the way they are talking with each other and they asked there's actually plenty of evidence now I can't remember the name but the chat pod that people actually fell in love with because you know they had interactions which were meaningful enough for them so I'm not suggesting we're at a point you know where they say is all sufficiently smart to make these types of relationships happening but we're actually getting there and I think there will be more and more of that and there's in fact examples in the healthcare space in Sweden where you use these chat pods for kids that start being depressed and they actually prefer these chat BOTS over human beings because I feel less judged so when you're depressed or of a mental issue you see most likely or less likely to go to another human being and explaining and sharing what type of you know your types of problems
and mental issues so talking to a chat bot that gives you sort of neutral advice is a much easier way to interact with a machine and get this type of advice so and then I'm seeing the time is going so my storytelling seems to be working but here on this slide I think what's very important which I'm missing here and under many other events is you know to have a ground a discussion of what AI actually is and I think a lot of people are still not completely clear they have been there are some excellent books which are usually learned when you start to study AI such as ruslan or weak artificial internal intelligence a modern approach and the point that I want to make here's or several points is that AI has been around for a very long time so it's not just the bus for the last 2 - 2 or 3 years AI has started in 19 probably around 1950 where Alan Turing brought a book called Ken machines thing and started
to introduce a Turing test so many of you are probably familiar with the Turing test which is a test of seeing whether any ice reached human level sort of performance in its intelligence and so that's one and then of course a few years later John McCarthy and a group of researchers pretty nerdy researchers right Marvin Minsky and a number of others started this whole field of AI which today has about 25 subfields so just like in medicine we have about 20 30 different subfields like a dermatology immunology or many others in artificial intelligence you have actually you have a very similar set up so people in this workshop back about seventy years ago started to define these different areas satellizer don't work okay but you know you see essentially these these different areas that are in red are considered to be subfields in artificial intelligence and they are well well as well-established they have been hundreds of thousands of papers published in this field over the last decades and there are many research and many
faculties that dedicate themselves to this type of topic so and then of course there's a question that comes up some people insist wanting to have a definition of AI and then I'm taking a step back and thinking often well why do you need a definition affair what for exactly right and so I think that it's a very important question you need to ask yourself in the medical sector that there is an answer to that so if you for example want to become a practicing practitioner so GP a doctor a physician there is a clear definition because you know you can't just not just anyone can become a doctor you have to go through training you have to be registered you have to be monitored it's a pretty rigorous field and I'm actually expecting something various very similar a similar thing happening in in AI and data science also because we're essentially dealing with people's life right it's people's life people's finances people people's human rights and not everyone should be tinkering with this type of
technology right there should be very specific registrations at the same time I'm warning against you know making it too over complicated which Europeans tend to do they make it very complicated and they look at Brussels and they look at Berlin and you know what type of regulation can be put up now what's in Brussels recently and I I came to the conclusion that well let's try to have as little regulation as possible as needed so I'm not saying don't have any and just go out wildly but organize that properly right so that that is something to consider quickly the this is just a reminder far beyond hinted to it and I think we go a little bit deeper into this big thing why suddenly everyone waking up to this you know why is a eye on everyone's in everyone's mouth you know and one of the reasons is the economic growth so it's about it's responsible for the majority of growth of the GDP globally this and then Gartner and many others have analyzed and
suggested that the it's the most disruptive class of technologies by far for the next 1020 years coming so then other technologies come but AI machine learning is number one many many name like three countries that really leaned into this the USA many years ago under Obama developed a really nice a thorough assessment of what AI can do as a multi-purpose technology which it is so it will be applied across every single industry and they met they wrote a very good report which is about 150 pages long I recommend you to read it is still very relevant so they leaned into this it hasn't been picked up quite from the with the current government another one to notice of course Finland I think so they have been a few years ago being the first to introduce a strategy for Finland in terms of AI and so that was very forward-leaning and I happen to live in Stockholm and I'm in Finland almost every week so I have a lot of interactions with with the ministries there
and with here with Prime Minister as well as in Sweden and in Oslo and Finland is very far forward-looking and so I also feel well why is it that Finland is doing it well it's in part because they are looking for new opportunities after some of those big companies not being as strong as they once were like Nokia and so on but also they hang I just say that on a private note I'm not you know I'm just it's my observation but they hang a bit there a go outside the door when you go into the negotiation room so the Prime Minister's a CEO I find them very grounded in the discussion and so that is actually very important prerequisite in order to make like very very strong-headed and good forward-looking decisions so the other one is of course jobs big big huge topic the WEF the World Economic Forum has estimated that in the next three years there will be 75 million jobs being lost and 133 new ones these types of estimate estimates
are really difficult to make because I mean no one really knows what a I and how a I influences things but it's sort of a ballpark figure so it is a net increase of jobs at 58 million types of you know 58 million jobs then at the same time I actually tried to try to distinguish a little bit more what I'm often saying is AI does not kill does not kill jobs it kills tasks right so because they are very good in solving certain specific tasks that require certain cognitive abilities so I think there's plenty more to do in the future then the other fact and we had plenty of discussions in the breaks I had at least quite a few on that topic which is the experts so wired and the New York Times have estimated that they're less than 10,000 people AI experts in the world that can truly move the needle that can really that really understand the business the technology the science and also have a good grasp on what it
means for society and in Sweden that's on if you statistically map it to Sweden that's about that's about five and in Finland it's about three so I haven't done the mapping for Germany but I don't think it's about maybe 2025 or something and so the other point to make is we we are nettle exporters in Europe altogether of talent in the AI space so far more people that we educating here go to the US and China now to then remain in in Europe and see that they have a chance to actually develop their skills which is a big warm you have to turn that around good if it isn't clear I think I will go over this slide reasonably quickly but I'm saying on what I'm seeing when I'm sometimes discussing this topic particularly in the healthcare space and the audience is full of physicians and see also of of big hospitals right so I mean if you're believing so one thing to think about is when when you develop products and services today anything
I mean even if this love of bread it requires many of those technology which technologies which have been groundbreaking over the last hundred years so there's almost nothing that you can develop today competitively without the help of electricity's or computers I mean the whole value change in supply change are relying on those types of technology and it's very similar with artificial intelligence so it's it's creeping in and I'm always also saying so I'm not a blind advocate of AI I'm always saying well if you have a business problem to solve try to I mean if you can try to solve it without AI machine learning and the reason I say that is it's very very complex and they are not enough people dealing with that properly so if you can try to avoid it but I think the reality is it's about the same type of message that I would say in 1998 we'll try to you know try to avoid using the Internet as much as you can't it will be an overwhelming wave
of technologies that that will be you know will be provided and delivered through through AI sort of technology and science so it can't be you can't be avoiding that all right Wow four minutes and I actually wanted to start talking about the the connection so what brought me into the field of AI and AI was I think one of the main reasons why we are intelligent and that certainly has motivated me and I studied psychology and social psychology and cognitive psychology is that one reason we have intelligence that we have effectively can communicate with each other and connect with each other so that that is one of those big things that I I'm a strong believer in and that drove in part my my my psychology thesis so when I did my psychology thesis I was focusing a lot on building trust between a human and an AI so I built at BOTS and that was a long time ago which were in the way it speaks to individuals would increase the trust in those
individuals in these humans it was sort of a real estate scenario and the way it would do it it would explain the decisions it takes in the process of this dialogue and so you could clearly see in our in our experimental groups that those that receive that type of information certain extroverts like these types of explanations so we measured whether or not they're extrovert and introverts and those that received more information actually took it as a positive effect those that are introverts were not not so positively affected but what when you map that to today so I see this red line like every industry and every company is now extremely interested in building this trust you know with the user so when you particularly when you have chat bots and they are starting to look at many factors so when you build a chat but some of you might think oh well that's sort of solved you know you go to some website today some bank and you have some chats well that's not even
we haven't even started that journey I mean in part you have you have things that you don't see like things like intention recognition so what you type in it may not necessarily be what you mean right so there's lots of AI software that starts detecting what you actually mean it's called intent recognition another important thing that is not completely correct and particularly in Europe with all these different languages is you know the languages so when you are when you're using sentiment analysis so were you after a chat happier or angrier as a resolution of the questions that you had to let's say it to the bank and then there's many other things like you can use biometric identification so are you actually the person that you say you are by the way you type in things just by the way you type in things you can detect a lot you know cause era and EDX using this technology to identify students for example when you log on so these are very important things then you
have multi modal interactions so when you start using and seeing people you know inter nations when you start talking to the bottom it recognizes whether you're stressed or unhappy entering conversations is a huge topic where by now most of us using these technologies just to say and give a command and say well you know turn off the light or something but maintaining a dialogue over many so-called dialogue turns you know takes takes a completely different technologies and not many chat pods are a people able to do that the other one is I did my master thesis a long time ago and again that draws me all the way to today's technologies and I've reapplied all my thoughts and my learnings from those days to what I'm doing now on all these companies but I was doing my master thesis on robotic soccer playing so at the time we had these robotic soccer dogs you know eyeballs and they were playing completely autonomously against another league so literally they were at first they had a picture
about the world model about the world they understood where other team members are right and they played as a team against another team and this is this is called this is one of those grown challenges which is posed to many AI researchers to achieve just like it was for chess chess playing in 1998 with Kasparov and also with the alphago 0 these are all games which when you achieve them it's considered to be a big milestone so this is one of those types of games that when they are achieved they are considered to be extremely useful for applications like search and rescue on hospital setting where a robot needs to for example lift a patient from one bed to another needs to collaborate with the different you know nurses and doctors in the hospital for examples and then of course you see many of these retail shops we built for example a system that goes through supermarket aisles and detects what's in the what's in these shelves and so it could would automatically sort of
reshelve them but at the same time also be available and offer information to the art to the visitors to this shop all right okay and then another one and that was that was a lot of my that stems originally from the PhD thesis whereby I looked at you know how do can you build autonomous systems and teams that would make decisions together and I was investigating under what conditions what these teams make optimal decisions so how would they be exchanging information with each with each other and aggregating information the most effective way so that the number one would need to exchange and talk so much with each other but still have enough knowledge about the world in order to understand it properly and so that's actually something that we started to use in in in the healthcare sector so when you have sort of care teams that need to make decisions with each other and so on so I think I have ah here we go so I think I just want to have two
more slides to go and one is well it turns a transforms value chains and work and I think one way to start looking at it is you have let's see yes yes so the way how I look at the tasks that are essentially that need to be done in order to create value and deliver these products and services that I mentioned earlier they're essentially three types of tasks which will be influenced one is you will be humans and employees will be augmented in the way they're delivering their share to their job to the company right and so augmentation would mean that as a radiologist or pathologists you would be you know you have all this data all these MRIs and x-rays coming in you see certain patterns in these x-rays but an AI is much better doing that sort of part and you know going and crunching all data together you will be much better able to make these types of decisions and that's actually now well-established also in the scientific literature that you know
a is are much better at pattern recognition but when you start throwing the knowledge and the wisdom of those let's say the biological systems and the artificial systems together you receive significantly better outcomes the other one is of course automation and now maybe in the background you saw maybe this type of video I find this particularly interesting because it shows how a a robot goes over fields and essentially sprays spray some weed killer on like wheat in the field right and it does it it's solar paneled it's a solar-powered and it goes back and forth and it has a deep learning algorithm in the front and it detects what's what's good plants and what are bad plants and it sprays them and in my world that's a very good idea of replacing these types of tasks because I don't know many people that want to build a career out of wheat plucking right so that's something where I think I see some very good use here and then of course there are plenty of new
tasks so I mentioned earlier when you with in the industry like when you have chat BOTS or other AIS you know you want to build AI set or that have certain personalities so many banks and many insurance companies the very important relationship that you have with the customer it is very important and you want that chat bot to represent the values of the company so you need to employ people that have the ability to put this personality into these types of chat BOTS and personalities so there but there are plenty of plenty of other new tasks and jobs that are coming so last slide I think the message and what I wanted to say is that the companies that will succeed and the industry and the research that needs to focus now on very important topics or the most important my opinion one of the most important topics is that trust and the collaboration with AI so trust in AI when you have that in your company and takes over certain tasks then you need
to build among employees a trust that that system is doing what it is meant to do what everyone says it would do and that's not an easy task it's a very difficult task in fact then the collaboration with the AI so you need to make this connection and the interaction the communication between EI eyes and the human being smooth and I'm not talking about robots are very sophisticated chatbots but also when you use predictive analytics and traffic control and so on because the system starts doing things that you didn't expect it to do and then also the thinking there's often this thinking and that will be my last two three sentences that you know this is one a I write we built this one AI system and it will be sitting there and it's doing all the analysis and then it uses a very nice prediction about where the market is going well no there will be very many AI so many will have an so-called agency they will have an understanding you know about
what's going on in the company and they need to align their goals and the way they are acting with all these other AI setter in the system as well as the humans and that's actually complete research area in its own right which has been around for 35 years so now many of these systems have become very very sophisticated and they they they are applied on a number of companies and I think those companies that take that to heart will be having a very nice leap forward in this new industry in this new era of artificial intelligence thank you okay so thank you very much there was good we have five minutes left for Q&A who is the question and would like to ask Christian a question you there you go I am Hagar from tu guys then just a very quick one so from my experience with a human way of behavior and human cognition I would say that like the opposite is going to be the problem right now maybe people don't trust a
eyes but we have we have observed that when the computer systems have been introduced in workplaces and workflows then people won't do things because the computer doesn't allow them to do it anymore so they kind of trust over thrust the computer against perhaps their own better judgment and don't you see that this is going to be a possibility with AI as well once the area has been shown to outperform humans in many tasks then the for example the radiologist is going to look at the single se or a is that and the ayah has been shown to be better than me so why would I just no trust the AI so you mean why so just to understand you mean why the AI system why you would trust the system automatically or yeah that the problem is that people don't trust the a and I think that is right now that is the problem being they're trusting it too much in about five years now say in some fields today is going to be sure
to be so good that the humans will actually trust the AI more than they trust themselves memorizing my tears are your thoughts yeah yeah and in some cases that's exactly right so well yes and so we already trust lots of technologies to to a good extent and if it's very reliable and it keeps you know it does deliver the outcomes that you expect it to deliver and it gives you the benefits and the value and the that the positive things that it should give you then you know I don't see it I don't see it as a problem and you correct that you know once we reach that stage where you see these you know tasks in a eyes doing jobs which you naturally can can see you know offering that benefit and that would be that would be good thing but at the moment just to reiterate the point I made earlier at the moment I see that the main parts in the workflows and the value change there is a fair bit to
do not on the algorithmic side you're not for the AI and the technology to do to to you know to improve but it's more like well if you bring it in and you have a sales executive that is you know looking at leads and needs to call our people to sell things for example any of this AI software that says well these are the five leads you should call up now and your Commission by the end of the year you know depends on whom you are calling up and where you spend your time and you need to have sort of a natural you need to have this natural connection to this type of you know technology and these types of outcomes but overall I see this as a much bigger problem at the moment than the other way around yeah I mean while Java maybe we can take this afterwards so I said it's a bigger question right so so but I I know I I think that but that's where it's naturally going there's
another maybe a quick response to just demonstrate where it's maybe not even the our choice of trusting an AI might go away Amazon a few years ago has actually launched a patent which would predict what you want to buy so the entire industry Amazon is essentially based on you know giving you recommendations it's a core part of their business model when you to Amazon you know you get recommendations and they there's thousands of top machine learning guys that are trying to optimize you know these recommendations so that you essentially buy four out of ten recommendations and then three out of ten recommendations becomes better and better and if it when it switches to a point where it's viable to send out let's say you know and this is by the way called the shop and then by a model and they started thinking about turning it around or making it as a shop and then ship model meaning that they predict who you are they sent you the item without you actually knowing what you
might want it arrives at your doorstep you think well what is that and then and that would be based on them being confident that that prediction would actually be correct right and so there's a plenty of implications one of it is well first of all first at all it takes away your choice in some sense you get these types of things and the second part is that those companies that apply these types of technologies will wipe out the entire competition because you essentially if the others are still well we expecting all these customers to go to on our website and shop and spend a lot of time well by that time you know Amazon and another company will have already sent these types of right so there's plenty of those types of things so I see more these macro type type impacts and individuals because they will see the benefits like with many other technologies