Dive into the heart of Sweden's AI scene with Episode 120 of the AIAW Podcast, featuring Jeanette Nilsson from Rise. This episode, "The Swedish AI Ecosystem," is a riveting journey into the challenges and intricacies of bias in AI, as well as Sweden's unique role in global digital innovation. We explore how AI reflects and shapes societal values, discussing the balance between cultural diversity and the risks of reinforcing dominant narratives. Get an insider's view of Sweden's digital strategy, the impact of GDPR, and the AI Act on businesses and public sectors. Jeanette's transition from civil engineering to AI at RISE highlights the transformative power of AI across various sectors. This episode is more than a conversation; it's an enlightening exploration of the broader impacts of AI, from small enterprises to large corporations, and a deep dive into the EU's legislative process. Tune in for a thought-provoking discussion that challenges us to consider our collective responsibility in shaping an AI-driven future.
Follow us on youtube: https://www.youtube.com/@aiawpodcast
Dive into the heart of Sweden's AI scene with Episode 120 of the AIAW Podcast, featuring Jeanette Nilsson from Rise. This episode, "The Swedish AI Ecosystem," is a riveting journey into the challenges and intricacies of bias in AI, as well as Sweden's unique role in global digital innovation. We explore how AI reflects and shapes societal values, discussing the balance between cultural diversity and the risks of reinforcing dominant narratives. Get an insider's view of Sweden's digital strategy, the impact of GDPR, and the AI Act on businesses and public sectors. Jeanette's transition from civil engineering to AI at RISE highlights the transformative power of AI across various sectors. This episode is more than a conversation; it's an enlightening exploration of the broader impacts of AI, from small enterprises to large corporations, and a deep dive into the EU's legislative process. Tune in for a thought-provoking discussion that challenges us to consider our collective responsibility in shaping an AI-driven future.
Follow us on youtube: https://www.youtube.com/@aiawpodcast
sense is a little bit provocative. You said at some point that you want bias in models. What did you mean with that?
Jeanette Nilsson:I definitely want bias in models, because something that sometimes frightens me is that people think that there is some kind of fairness and divine in the name of artificial intelligence. They don't understand that it's a mathematical method and it doesn't get better than the things that you put in. You think that if you don't understand also that, this basis on values, it's like if you have every person in the world as one data point and then you think of now we have fairness in the world. Sweden, we are 0.1% people. In Sweden. We think that every person has the same value, whether it's man, woman, ethnicity everyone has the same value. Girls go to school, sweden votes, women in Sweden have voting rights and if you kill a woman in Sweden, it's a crime. If you look at these people mostly in the world, it's not the same fact. So if we don't have bias in models, we just look at the numbers, then a woman will not have voting rights in the future. So I'm definitely for bias.
Henrik Göthberg:So the core topic here is that how can we work with AI in such a way that it reflects the cultural values of many different countries and that ultimately, you would be understood as bias if you look at them mathematically on the total population of the world. Is that a summary?
Jeanette Nilsson:It's a good summary because it's interesting that when you have data, then you don't think of it as bias. But if you see a kind of data and where you got the data from, then you may be okay. This reflects my reality. And if you don't have understanding that cultures are different and people have different kind of both experience and possibilities, then you can think, oh, this is not bias, this is the color of skin for me, but I'm a really small percentage of people. So I mean, I think that's super important to think that this is not going to be like one size fits all. We have to think of different cultures because we are so alike.
Henrik Göthberg:But do you get a lot of pushback? Because that is a super provocative statement. But when you unpack it, there is a lot of truth in here. And I give another example. We had Erik Hugo as a guest here last week. He's born in South Africa. He's born in the same hospital as Elon Musk, like two months away, so he's born in this era where he grew up with apartheid, so he has this whole feeling. He's been working in Silicon Valley. He's working at the Silicon Valley company right now and he's Swedish since 25 years and he loves Sweden. He loves our values.
Henrik Göthberg:And he really taught me a lot about the nuances of language and what happens when everybody simply speaks English or the model, even if they understand Swedish. The whole model takes your Swedish language into the large language model in English and then back again and it actually spits out something that culture is a little bit different. And to me this is really interesting because when we say bias, it has such a negative connotation right now. But if we maybe put another label on it, it's about finding the values and the cultural diversity. And how can we talk about the truth, about cultural diversity, without being pigeonholed as bias? Because is it a bug or is it a feature?
Jesper Fredriksson:And it's also a technical question and it's a decision on how to steer the model. So after you've done the pre-training on a big set of data usually the whole of internet and then you do some kind of fine-tuning of the model, and here we could go into the Gemini the buckle that happened.
Henrik Göthberg:Yes, this is a good example of this right, could you?
Jesper Fredriksson:explain. So then there were a lot of pictures on the web where people tried to generate different people from different eras of historic times. So, for example, they tried to generate Nazi people, but the images turned out to be Asian people in uniforms, and there's a lot of examples in that line of thinking. But it seems to be hard for Gemini to create white people, basically because it's been fine-tuned in a way to support, in some way, diversity, or Google's version of diversity, would the critics say? So they got a lot of pushback from being too woke or whatever you want to say.
Henrik Göthberg:If you're Elon Musk, then you're criticising a lot, Then you would think Google, like some Gemini, went woke. That was the joke.
Jesper Fredriksson:So of course he made a big thing out of this. But the question is if you don't do anything, what do you end up with then, and should you steer the model, or how should you steer the model?
Henrik Göthberg:Are you provoked by Gennett's comment on S&EI engineer? A little bit, let's debate. A little bit, let's debate now.
Jesper Fredriksson:So if I take the other side, I can definitely see the side of the engineers at Google trying to be fair in how they want their model to be seen by the world, and I'm sure that somebody was thinking good thoughts and trying to. Let's not only portray white people. Let's make sure that we have other nationalities, other culture.
Henrik Göthberg:I think if I would steal man or argument against you. Of course people want to do this in cultures. If we take the Swedish culture as an example, we have maybe extreme equality, a whole welfare system, the way we have children, the way we have our child support. Actually, you're really pushed, as a woman, out to work even if you don't want to. I lived in Australia for several years and then there's no system where you can have daycare subsidized, so there daycare for children is $100 per day. So literally it's not someone is home with the child. If he's the guy or if the girl, doesn't matter, but someone is home. It builds a whole other narrative.
Jeanette Nilsson:It's a completely different narrative.
Henrik Göthberg:And when we then the Swedes trying to explain our narrative in Australia, it doesn't fly. They can't really connect with it. On the other hand, my wife told me like when she came back and she was really immersed in the Australian narrative, it was almost like, oh my God, you're home with the kids. That's really strange. No, no, no, no. You need to work and in the end we want to. I think I'm from the point where you want to decide this on your own right and there are different views.
Henrik Göthberg:and how then to view models? Should models streamline everybody into one bucket, every, everything should look the same. Or how can we have cultural diversity, how can we have diversity in values without ending up in this sort of war zone of vocism or not focus them and all that? It's really strange, it's really difficult, I guess. And who gets to decide?
Jeanette Nilsson:Yeah, it feels like now we know who gets to decide a lot of things, since open AI put out all out these models. I have a colleague that is working as a researcher and she made a presentation the other day for group and, and, and it was on bias, because I said because she was also a bit broke of that, but then she showed me that she had made trying to make pictures of CIOs, and and in she had information officers. Yeah, yeah, yeah, the no, no, yeah. The executive officer, the highest CEO.
Henrik Göthberg:Yeah.
Jeanette Nilsson:Yeah, and so she had prompted differently and she got pictures and and they were. She showed 10 of these. It looked like the same guy white, 40 ish, handsome.
Henrik Göthberg:This is it.
Jeanette Nilsson:So everyone looked the same. It was a bit of anger, but but everyone looked the same. And then she asked the Dalio also to produce. I've heard I don't know if this one is a myth or nothing or not it's someone told me. When Shad G P T first came out, someone asked him what is the most common profession for a woman? Have you heard that?
Henrik Göthberg:No, I haven't heard this.
Jeanette Nilsson:Okay, so I don't know if this is true or not, but and the answer was prostitute.
Henrik Göthberg:It's the oldest job because of all images of women on internet.
Jeanette Nilsson:Most of them are poor.
Henrik Göthberg:Poor, poor, and the so so and and the so I don't know if that's true, but my my research fellow, she.
Jeanette Nilsson:She asked about the picture of work and men, work and women, and it came up with 10 images and in the 10 of men they were kind of powerful poses and so on. One was naked. In women it was kind of not powerful poses and only two had clothes on.
Henrik Göthberg:So even though, if you're so, then then you now you're still manning your own argument run around bias, right, because then we have different challenges around how to deal with this, right?
Jeanette Nilsson:It is, I mean so. So for me, bias is that we have to say what kind of values do we want to have? That's bias, right.
Jesper Fredriksson:Yeah, so I guess the engineers at Google tried to do that. It's just that they didn't do enough of a good job to do it.
Henrik Göthberg:But is this even solvable or does it do become? What scares me is when we end up in a way that there is one size fits all for all of the world, which will not work, and I kind of like when Elon Musk came up with grok, the large language model obviously it's going to come in, going to be roasting people be rude and all that.
Henrik Göthberg:It's a little bit like we need to get to the point where you can choose which AI personality or values that suits you, and I have another argument on that. This is why it's super important that there are language models that are sort of Swedish to really connect with the Swedish values, and that there is then there is done, real value of having that tricky now to you know, and then we can get into the debate. Should you do a Swedish language model in an American model? Is that good? Or should we rather do a Swedish language model in maybe an open source model from Europe? You know, because as soon as you go into the American language model, that will influence the cultural values, if I believe in Hugo.
Henrik Göthberg:And I truly believe it. The Swedish language and culture I think has it sort of different languages will take you in different directions, like some are more philosophical, some are more commercial.
Jesper Fredriksson:I would argue have you heard any European projects on specific language models. I think that there's, for example, finnish language is very special, so it's hard to find a good language model.
Henrik Göthberg:Do you know the Solita poro?
Jesper Fredriksson:Yeah, exactly.
Henrik Göthberg:This is one of the ones. I know that's where it was going, right.
Jesper Fredriksson:So I wonder if it's that a common thing in like other examples of having.
Jeanette Nilsson:I think everyone wants to have a model in there. The most interesting thing that I'm kind of following a person I'm working with in from Italy they're trying to do a model, a language models, on culture.
Henrik Göthberg:Wow.
Jeanette Nilsson:Yeah, I think that was interesting because that is such a big part of us and I'm thinking also about we mean different things when we use different words. Like you're saying sorry, like an, even within Sweden, I mean yes, of course, if you say something in the northern Sweden it has a totally different context than in here in Stockholm or maybe in Skåne.
Henrik Göthberg:It's so true. I grew up in Buruos in and then moved to Stockholm 99, living a couple of years in Australia and between there.
Henrik Göthberg:And those things is really I mean like it's different compartments in my brain. And someone even said, like Eric said, it's like he said really well, South Africa is a super. He grew up in a very, quite violent place. He's sort of you know, if you read the Elon Musk biography or it's, it's like, oh, he's not unique, everybody had that upbringing in South Africa. So he said he realizes that he, when he talks South African, he has a much more aggressive vocabulary. So actually he is a person when he switches between Swedish and he goes and talks like a sort like he goes into his South African brain, sort of speaking. It actually makes him be more differently personality and I totally believe this, Definitely yeah definitely, I think, language.
Jesper Fredriksson:That's my experience. When I started speaking English professionally, I felt like I went into different personality. When I started to speak English, like if you go abroad, the first times I did it, I invented a new personality because I was a different person. I'm in a different country, I'm doing different things, I'm speaking a different language. So I felt that do feel the same.
Jeanette Nilsson:And especially if you want to.
Jeanette Nilsson:I mean jokes or something else like connect it's hard to do if it's not in your kind of mother tongue, definitely differences. And I'm thinking sometimes that we as people I mean if we see all the evolution we as always kind of we are so lazy, we want something to do for us. I mean, in the beginning you had to be really strong to be to survive, but now people don't have to be strong anymore. I mean that has also some negative effects on our health because since we don't work out, we don't do things like that. So I'm also thinking of if we're the prolonging of this curve that we say we want to have so much artificial intelligence. People don't need to learn to compute, they don't need to learn to count, they don't need to learn to read or write or anything. And then I feel like where?
Henrik Göthberg:are we going?
Jeanette Nilsson:Yeah, what do we want? We want to lie in bed and look up in the ceiling and be part of meta-words.
Henrik Göthberg:Yeah, it's the Wally movie. The Wally movie, the old Pixar movie, where they all floating around in a super fat.
Jesper Fredriksson:But with this I think we have that as the last point. That's the last point. Super fat yeah.
Henrik Göthberg:But I think it was an interesting introduction. But let's do the proper introduction. Jeanette Nilsson from Rise, super excited to have you here.
Jeanette Nilsson:Thank you.
Henrik Göthberg:I am actually one of the key things I'm excited about. We don't know each other from before, but I already picked up on it. How do you make decisions happen in EU? We need to talk about that. I mean, how do you make decisions happen? You have a background, you work with EU. You can introduce yourself in detail. I also want to introduce the theme for today, before we get stuck into who you are, jeanette. The overarching theme is to talk about the AI ecosystem. I think you framed it even better here when you got here today. It's like how do we look at this Swedish AI ecosystem and how do we together become a strong force with a clear direction and clear focus? Together, in order to make things happen, in order to have a voice as an example in EU, we need to really work with a focus or something. Now we have a good, great ecosystem, but how can we make it focused, I think can I summarize the theme?
Jeanette Nilsson:in that way.
Henrik Göthberg:Yes, perfect. So, with that theme in mind, welcome Jeanette Nilsson. And also, of course, anders is not with us today, so we have Jesper, hello, and Jesper, what's your surname? Fredrikson? Fredrikson, so I was guessing between. I was thinking Jesper Nilsson, no, no, no, that was Janette Nilsson. So Jesper Fredrikson who is a good friend of us, and he's been on the pod before, so you can check out Jesper as well, and you are working at Volvo Cars True.
Jesper Fredriksson:As an AI engineer, as an.
Henrik Göthberg:AI engineer. So I'm the sort of the business guy and Anders is the techie guy. So now you're standing in for the techie guy.
Jesper Fredriksson:Yes, I'll try to fill his shoes. It's hard but I'll try.
Henrik Göthberg:That's great. But with that, jeanette Nilsson, could you please just introduce yourself a little bit what you do in DRISE and your background, and then we can move on from there.
Jeanette Nilsson:Right. So I'm Jeanette Nilsson. I'm born and I'm living in Luleå and I've been working a lot of places, always troubling. I wanted to be a ballet dancer.
Henrik Göthberg:Yes, that's what you said. You introduce yourself as a ballet dancer. Yes, sometimes I do actually.
Jeanette Nilsson:And then, because that was my life, when I was young and my father said you know, jeanette, I think it's better that you become a master of science instead. And I listened to my father. I started at the University of Luleå, technical University, ltu, yeah. And then I started to work for the road administration, building roads and bridges, and I was the only woman in 800 people.
Jesper Fredriksson:An actual civil engineer.
Henrik Göthberg:Yes, a proper civil engineer and I was the real project, the big ones.
Jeanette Nilsson:Yeah, yeah, actually there is a bridge still that they call Jeanette's Bridge, and I was 25 years old when they made me the head of this kind of building, this bridge, and it was such all my male friends from the university like, wow, are you already chief of a building of a bridge?
Henrik Göthberg:So it's Jeanette's Bridge? It's Jeanette's Bridge. Where can we find Jeanette's Bridge?
Jeanette Nilsson:It's in Pithio. Actually, pithio, which one is it? Yeah, they have some. You have to be in Pithio to know where it is. So that's cool and that was really nice. But then I wanted to work more on designing things, not building things that other one has decided, and we were part, then entered in the EU and I also wanted to work internationally. So then I started to be the EU person for the road administration and I got in trouble with the Narnist department.
Henrik Göthberg:Actually the Narnist department. Yeah, yeah.
Jeanette Nilsson:Because I went down to Brussels and I said, oh, we want to have these roads in the national agenda and so on, and they said, oh, you must silence this woman.
Henrik Göthberg:She's kind of disturbing all our negotiations.
Jeanette Nilsson:Please keep her away from Brussels. Yeah, so, and my boss, he said ah, I just tell you, but go on, go go, go, go go.
Henrik Göthberg:So he was smiling yes, yes, go, go go.
Jeanette Nilsson:Yeah, yeah. So and then after that they recruited me in Brussels to work with the firstly infrastructure and then other things, and I learned actually the real and I have realized that I actually want a few people that have kind of discovered the loop, because most people do the same mistakes as I did in the beginning, talking to wrong people, and how to get shit done in you. How to get shit done in you.
Henrik Göthberg:You can probably write the book for that.
Jeanette Nilsson:You for dummies and make a backload of money on this and that's something that I'm trying to use now in my work, because after that I was working in as the in a have working, working with as a business development Narnistly chef, it's called in different municipalities in Norbotan and also started up a gaming company community. Oh yeah, and and that was kind of been forced 2016, something like that Arctic Game Lab in and I was Arctic in lab.
Jeanette Nilsson:It was five municipalities Shaliff you had started to have a gaming education program from LTU and they had a lot of people, smart people, a lot of people that are working in gaming industry, often than other parts of Sweden and so on, so it felt like everyone doesn't have to move to Stockholm to do this. So then we started to to have a kind of ecosystem in the north, in five municipalities, and in the beginning I was the shareperson for this and it was really nice because they felt like, oh, a middle-aged woman so good that you kind of talk to people about what we boys are doing so.
Jeanette Nilsson:The boys, the cool boys, the gamers, yeah, got the you as their hero, yeah, and as a spokesperson, the broker, to the and because they have felt for so long time that people had kind of put them in a kind of position where they were kind of not so good, kind of bad, instead of one.
Henrik Göthberg:Seeing the super opportunity yeah.
Jeanette Nilsson:Technology and the way how to reach out to people, and especially if you see if you are gaming in doing a game, it has to be good graphics, it has to be interesting. If you are doing an instruction video for someone to kind of manage a data center, it can be super boring and super bad quality.
Jeanette Nilsson:So I'm thinking, when I try virtual reality first time, I felt like, oh, this is how I'm going to work with my mother and I feel like here in that Sweden was the best, finland was second best. But then Finland, they put in a lot of money to attract people from all over come to Finland and be in this level up kind of positions and they let people come and they gave them money and to develop companies and so on.
Henrik Göthberg:So Finland kind of but this is one topic we can talk about later, because you've seen this first time in Brussels how the Swedes work and how the Finns work and how they do a really good job in your opinion and how they have sort of pushed the AI again really well. But how did you end up in RISE?
Jeanette Nilsson:Yeah, I was recruited to RISE to kind of connect the ecosystem that I've been working with in Sweden and also in Europe with the researcher RISE, and I had the what was the role you were recruited into? It was like AI ecosystem driver. Ai ecosystem driver To drive up an ecosystem in Sweden and also connected. Now I'm calling myself AI ecosystem expert.
Henrik Göthberg:But Anders, who you knew from before. You guys met at the AI agenda, so is this the same time?
Jeanette Nilsson:It's the same time yeah, because I was recruited to that. And then Sweden this was 2019 and Sweden a lot of people started to have agendas for in AI and most of them have done that purely on research and RISE. That's, we are a state owned company, so we can't get direct assignment from the government. But we have talks, our CEO have talks to the ministers, but they can't like say you have to do this and we don't have any regulating spray or something like that. We have. We have money from the government and they kind of in the research bill we have some kind of assignments.
Henrik Göthberg:But help us here. Could you frame the context of the role AI ecosystem expert, you call it? Now. So what's the what's the regulating spray? What's the objective of the role, so to speak, and what does it entail?
Jeanette Nilsson:Well, it has nothing to do with regulating spray. So yeah, yeah, yeah, so no no, but my role is to now, when I've been working with this for five years. When I started and I met Anders, we were building up something in the AI agenda, and that was that's something that differs from everyone else in the world, because in that, when they asked me if I wanted to be the leader, we had a good a lot of experts in the in rise, and especially Daniel Jellblad, who was our expert.
Henrik Göthberg:Daniel, of course he's been on the board. Yeah, yeah great.
Jeanette Nilsson:So we two worked together in the beginning a lot with this. So he was the expert and I was kind of driving the a lot. So I took in other people, other people's perspective. So in this the work that we did, it was really from the grassroots level. But and not and we decided not to try to do something to. Here is what we want. Please give us money.
Henrik Göthberg:How did you understand when you put the group together? How did you frame your goal or your mission at that point?
Jeanette Nilsson:The thing that we wanted to do was kind of come forward to the minister with the suggestions for Sweden to kind of start working more with AI.
Henrik Göthberg:And what year is this?
Jeanette Nilsson:This 2019. And the idea was that we would present it in in the beginning of 2020. When COVID came and a lot of people they just wanted to be part of it, it felt like, okay, we want to have an Apollo project in Sweden. So people in Sweden really, really want to do things. Sometimes we talk so much bad about Sweden, so it feels like sometimes people think people from abroad sometimes ask me do you even like Sweden? You're living there, but you even like Sweden? But this shows really the contrary. People put they didn't get any money for this, they just wanted to work at it. So we had six different working groups and put in the end we had 25 suggestions and a lot of materials on what we what different people, different perspectives, experts and newbies.
Jesper Fredriksson:Do you have any favorite projects from that time?
Jeanette Nilsson:We have 25 different things suggestions that are ongoing, so none of them are really kind of finished because they are big ones.
Henrik Göthberg:It's big focus areas.
Jeanette Nilsson:Yeah, it's actually. They are on the. We have a website for this also and actually people are downloading it and I don't know really how they know that there is there, because it's kind of difficult to find out. So people still. I mean, we presented it 2021 together with the minister then, and the idea was that we will do things together. We will try to instead of. I mean, the problem, a lot of the problem with the ecosystem in Sweden is that Sweden has this tradition of not being kind of like.
Henrik Göthberg:In Finland, they say this is the way or the highway, finland has to adapt to this and then basically they get to take, they can create a force, because everybody's signing up to that one message and we are not so much.
Jeanette Nilsson:No, we don't do that and like, especially if you see all the authorities, they get a little bit of money, and a little bit of money in Finland. If you take Finland, for example, there are other countries but Finland is good because it's close and it's only half the population as we are. So we can't say oh, there are so many because there are so few, but still they are more present in different deciding kind of structures that are in Europe. And I'm thinking it surprises me a lot that we have been together with Finland. We became members of the European Union 95. And that is almost 30 years. It's more than 30 years ago and still it's like something we are still learning as we do.
Henrik Göthberg:And they have perfected its way? Oh not, but they have come a lot further.
Jeanette Nilsson:Yeah, they're kind of seeing OK, this is the investments that we are doing together and this is our common play field. We are putting a lot of since the I mean, especially if we are talking about AI and digital kind of competition it's not. It's not on geographical site, it can be anywhere. So it's a possibility for everyone, but it's also a threat. If you don't understand that, then sometimes if you so, in Europe we have to work a lot more together, and I see that a lot in my work. Germany and France they are not shy and a bit an Italy to say this must, this must be good for our industry.
Jesper Fredriksson:So you're saying in Sweden, we should align more internally and then be more aggressive, more of yeah, yeah, yeah, yeah.
Jeanette Nilsson:And the thing I mean we are always part of every decision in this in Europe and we decided for the first time in history, all of the three entities the European Council, the European Commission and the European Parliament had decided we are going to work for the digital decade, this is the future. We and they put all the efforts to try and to gather all the policies, all the programs towards digital ingredient. This is the way forward for Europe. We want to do this and we're going to have everyone on board SMEs and so on, everyone on board for this. Then, if we don't understand that this is where we are going to put all the common efforts, people from different countries, different. We have to think that what Germany does we have to be aware of, because it's also a bit of internal competition. So it feels like in if we like Finland, when they see that in five years time we would like to have this position, then they start it.
Henrik Göthberg:Yes, this goes back all the way to the conversation we had before the pod, and I think this is a really interesting topic in its own. So we're going from the AI agenda and we are slipping into how Finland and Sweden is different and how it really works in EU around these topics to get focused, so please carry on. I just wanted to sort of the context is sort of moving into the EU perspective here, I think it's very interesting, let's go there.
Jeanette Nilsson:And because more and more we have to work more together in Europe, because the competition is from the United States and it's from from the Asian part, and if development always happens somewhere else than in Europe, then we are going to be subcontractors to everyone else.
Jeanette Nilsson:And subcontractors don't make money. And in Europe we have this way of someone said to me people in United States they want to work, they are never vacationing. In Europe we won't have vacation, we work for having vacations. So if I think the important thing and the ecosystem in Sweden, if we could align, like more like people in Finland doing, to try to see OK, these are the three or five things that we want to be best in. For example, everyone thinks that Sweden is best in large language modeling. But if we don't do anything kind of fast to kind of nail this position, Finland will have it. And it's because if I don't think that Finland everyone's to kind of train as a foreign current, but if we don't do anything, if we think that, OK, this is our position by standing still today or moving slow means slipping.
Henrik Göthberg:It really means slipping If you're not following the productivity frontier, if you're not following the acceleration. Here you think status quo is OK. Status quo means slipping. Is that fair?
Jeanette Nilsson:Yeah, definitely so. It's like they're standing on this kind of transport man and you're moving backwards.
Jeanette Nilsson:And that's a sad thing to see that in all kind of different kind of measurements or you can have different kind of in all of them Sweden is a kind of losing position, such, and in some for, especially for the growth in the country, sweden is in the last position, europe. So I mean, I think, the important thing for Sweden, since this kind of technology shift is very crucial. We are good in the consumer perspective to do this, people in their everyday life.
Henrik Göthberg:We're very tech service, individuals, consumers.
Jeanette Nilsson:But we also have to understand that if we want to have this welfare society, we have to do a lot of changes, a lot of things more smarter, efficient. But if you say that people like it, no, I'm going to lose my job, instead of thinking, if I can do this more effectively, I can do all of these other things, trying to because Sweden has to step up. It's not like, yeah, I can just sit here and someone else will do the work. Now, if that happens, then everyone else will come, because now I think it feels like all people know and especially with war and everything, that if you don't do, if you don't move forward, someone else will be influenced.
Jesper Fredriksson:Do you have any insights on where are we right now in Sweden If you compare both within Europe and within the world? Where, on a scale would you say, we are?
Jeanette Nilsson:If I just say that what they asked Daniel from Microsoft that in a meeting.
Henrik Göthberg:Ekinne yeah, ekinne yeah.
Jeanette Nilsson:I have hard times with that.
Henrik Göthberg:Yeah, hard times with the last thing.
Jeanette Nilsson:And he said well, you know, europe is number three and in Europe Swedish on the lower half, Wow.
Henrik Göthberg:So he says he is a deep, knowledgeable guy in tech, deeply connected in Microsoft. So he actually has a picture on this, I think better than most, and he actually puts us on the lower half in Europe right now. And what is the framing or definition that puts us on the lower half? Is the investment, is it the focus, is it the know-how, is the competencies? You know, or do you know what he means with the lower half? I believe? It. I'm just curious in what he sort of puts in the scale.
Jeanette Nilsson:Yeah, I don't know what he puts, but if I see, you do it. Yeah, if I see, then in Sweden we have something that no one else has, is the VASP money.
Henrik Göthberg:Yeah, fantastic.
Jeanette Nilsson:And that also shows that the national amount of research money are really low in Sweden.
Henrik Göthberg:So that's one thing that takes so if you take away VASP, it's ridiculously small amount of investment from the state. Stupidly small.
Jeanette Nilsson:Yeah, super.
Henrik Göthberg:So it's almost like they got lazy and they think VASP is going to save them. It's almost, I guess almost super angry with this.
Jeanette Nilsson:So that is one thing, and then also a lot of other countries they are doing, they are using. It feels like we are in one way in Sweden very scared of leaving out our data, but we have our income, we have everything, but we are.
Henrik Göthberg:It is so strange and we should explain this for listeners who are not from Sweden that we are super scared about the whole thing. Data topic At the same time, anyone in Sweden can go out and go to a website and put in a name and get a very lot of detailed information about who you are. All the stuff that is called sensitive data and GDPR for some reason works to get hold of. I really don't understand how that can be compliant by the way, let me take the other side of the argument.
Jesper Fredriksson:Are we really scared about data in Sweden? That's not how I feel about it. That may be where I'm sitting.
Henrik Göthberg:No comparing what we are giving out at allaBullockse. If you compare that to Germany, germany who has this? They have a background that we need to understand with the East and West Germany and all that they are. I worked with Waffenfaden. I had staff in Sweden and Germany and Netherlands and it's huge difference in very, very big difference.
Jesper Fredriksson:In what aspects are we scared about giving out data?
Jeanette Nilsson:I'm thinking one of the. I meet a lot of people, both companies and some in the public sector, that first were afraid of GDPR and say that we can't do anything anymore because of this. Now they are super afraid of the AI act and all the regulation. The fear is that they will be fine, that they will have to pay a lot of money. I'm thinking also that in one way it is for protection of people, but if we can't solve this, then people I mean look at chatGPT. People know that it's a crazy. It can give complete false information and can steal information from patent and information, but people still use it.
Henrik Göthberg:But let me try to answer your question because I think this is a very interesting question. I think that the individual, the persons in Sweden, are actually less scared about managing their data than most countries in Europe, way more casual with this view than Germany, way more as an individual.
Henrik Göthberg:However the companies and especially the public sector. They went on the biggest scare. It's like being a huge scare in the Swedish public sector. We can't use cloud, we can't use this, we can't use this to the point where they're bending backwards on really idiotic topics in a way that is not even logical. So the whole debate in public sector. I decided not to work with the public sector because I think they're mad.
Jesper Fredriksson:I think they're literally mad when it comes to public sector that I can't speak for, but I believe what you're saying. In companies that I've seen, I think that people are realistic about GDPR, I think.
Henrik Göthberg:It's a mindset. I'm super impressed. I worked with Wattenfall when GDPR came. Super impressed, first of all, got real expertise as lawyers in early. Of course they can do that. Wattenfall is one of the big utilities. And then hardcore, instead of being scared, understanding how it works and basically say the core topic what is legitimate interest? Very simple If you want to have an electricity bill, I need to be able to have your data. If you want to have a new campaign, I need to be able to send marketing to you. And then they nullified the core topic by really being pragmatic on how they looked at this. And then super good education internally and a lot of discussion on it and basically this was never a problem in.
Henrik Göthberg:Wattenfall.
Jesper Fredriksson:Hardcore. That's my experience as well. Super smooth. A lot of work, but super smooth.
Henrik Göthberg:But if I contrast that to other companies who went scaremongering on instead of going head on into it and just structuring it, doing a lot of work but instead of getting scared, I think there was a lot of organizations and public safety that simply got scared. I don't know.
Jesper Fredriksson:This is a rabbit hole, but it's interesting. I've always seen like there's always some legal person at a company that takes the lead on this and says this is how you should think. And then it's like, yes, you can understand the thinking and once you internalize that, then GDPR isn't such a big deal.
Henrik Göthberg:The big deal is when you stay on an abstract level and don't really learn. As you said it before, we have a problem where we stay abstract and distant. Yeah, nothing to concrete. But you said it better before? How did you phrase it? When we were upstairs? I don't remember that. No, sorry.
Jeanette Nilsson:Well, come on. No, no, no, that's true. And now we have the next step, and it's the AI Act. I mean, we are going to use data information because people want to use it, and it's going to be in everything in the future. So I think that the important thing is to really understand okay, I'm going to do a device that helps people to wake up in the morning, or I'm going to do a device that everything is going to collect data.
Jeanette Nilsson:So you have to just think of that when you do that in the beginning. But the problem is, when you are here, that it's going to be a fine on 50 million euros if you do anything wrong. Then it feels like better, let someone else do this wrong before I do anything. So I'm thinking that is and we could have begun a long time ago in Sweden also to kind of prepare and see okay, we are the best country in Europe with public administration because we are really good. I don't know why we don't have any confidence in that, because we are. I mean to give something in a spectrum in the end and to tell someone from another country how we do our declaration.
Jeanette Nilsson:They are kind of it's like they don't believe it. So and we have a lot of things like if your kid is sick and you can just kind of put it into your computer saying, okay, I'm sick, my child is sick, and you've got money in your bank account tomorrow. So we have a lot of good things, but one of the I think, of the scare things I have friends that have children in school and they can't get out the list of pupils Because if they have to phone someone, they don't know who to phone because there are no lists for if the kids are going to something together.
Jesper Fredriksson:Yeah, that's probably to your point of public sector, it's more challenging but it's scary.
Henrik Göthberg:But help me out here because I want to structure a couple. I think there's a couple of very interesting topics that is related, that leads into each other. I think we need to spend some structured time on. I want to explore together with you. First question what is the context when we say the AI ecosystem?
Henrik Göthberg:And, being an expert in the Swedish AI ecosystem, I want to hear your definition and view of what we mean with the ecosystem. This is one. The other one I really want to get back to. I want to have the storytelling anecdote I think I think was brilliant use when we met how you were the tips and tricks making EU decisions work in your favor for dummies. I think that's a beautiful anecdote, what you thought it was and then how it really works. Because I think this is telling a lot about what we need to do in Sweden, how we need to think differently in Sweden and this. So I really want to get back to that topic on how do we get decisions done as a task towards getting focus and all that. That is another super interesting question. And then we have a couple of bigger topics like coming up. Now we have the Swedish AI Commission, what is really the Swedish approach to AI and all that.
Henrik Göthberg:But could we take it a little bit like structured so, let's start with what is the context of the AI ecosystem use? We put that to bed. How do we understand the Swedish when you have that title now in rice and all that? How do you frame that?
Jeanette Nilsson:I have decided. Title myself. Because I think I'm an expert. Yes, well, in Sweden we have so many different entities working with AI. We have the universities that work on certain level, but not all of the disciplines in the university. I'm thinking one of the problems that I feel that is that people don't understand that this is something that kind of concerns everyone because it's going to be in everyone's daily life for the rest of life in the future. So in the Swedish ecosystem there are different kind of systems that are working.
Henrik Göthberg:One is the universities, the academia.
Jeanette Nilsson:then we have some of the public authorities working together, Skatteverket and others that are for tracking, and they are really ahead of others.
Henrik Göthberg:We have a good friend, patrick Ekimu they are running in Bullock circuit and they are running even some EU projects.
Jeanette Nilsson:They are really doing a lot of good things, and he could also be part of the second discussion on how to work. So these are different.
Henrik Göthberg:Then we have this so we have the academia as one. How would you frame? We have the major verken, the general authorities. The public authorities the authorities that are taking a lead. That is actually a little bit ahead in terms of how to deal with AI Skatteverket is actually really advanced in. Sweden. We have a couple of examples like that.
Jeanette Nilsson:And then I feel like many companies bigger companies are working not kind of connected but in different ways, and some of them here. I think that it could be something even better. But if I'm trying to first talk about different areas, then we have now starting. I mean, skatteverket are now starting to see that this is in. We want to change. That was one of the 25 points that we had. We want to change the education system for this and municipalities are starting to understand. Also, we have to do something about social security.
Henrik Göthberg:And you can really understand this also from a. You know, I did see different consultants in focus on different parts. So we have the academia is one, the big authorities is one, we have a municipality movement going on and then you have an educational movement going on.
Jeanette Nilsson:Yeah.
Henrik Göthberg:Okay, this is all public sector.
Jeanette Nilsson:Can we, can we? And also a civil society civil society.
Jeanette Nilsson:Yeah, they had. They are also trying to to see that how is this going to be influencers or effectors or or other? So I've been to meetings with that because that was also part of the AI agenda, to see how people that are having kind of disabilities or others, how can, how can this be a tool that makes people more equal? But then when you're starting to work, big industries are always like you say what about big industries? They, they have a plan, they have money, they can do things. And a lot of SMEs, small companies in Sweden they are subcontractors to these big industries and these big industries are really important for Sweden. Most of these SMEs, they have more of a trouble, more to see. I've been talking to these kind of branch organizations and so on. Tech Sweden. How can we help these smaller ones? And here has the European Union trying to do initiatives because they see also that Europe can miss, can, can stay in some almost only small companies.
Henrik Göthberg:And this is not now. You're getting close to the heart of Teradax. So I grew up working in large enterprise around data and AI and I think they are. I see a very clear AI divide emerging inside Sweden in terms of enterprise and organizations, where some companies of course had the money and had the data and it came early on the agenda, like Wattenfald 2012. I've been working on this topic in Wattenfald and my goal has been with Teradax. What happens with the mid-sized market? I'm not even talking about the small companies. I'm talking about very nice Swedish companies with one to five billion-seq turnover Nice companies. They have not even started. This is, to me, the scary part and this is what we were talking about. How do we get to that segment?
Jeanette Nilsson:Yeah, and I meet that segment quite a lot. One of the problems also when I'm meeting smaller companies I mean up to 50 people is that most of them have their data in their heads and their hands and actually I have to talk about what data is and if they can actually do a lot of things with just having bills in AFFECTUR and simple things in administration. That would really really help them. And then I meet another group of small companies that are super skilled, often coming from universities, and they are working with the smart AI models, but they don't have any customers. So when I'm trying to ask them what we can do in the AFFECTUR system for them, they say, well, please help these big, for example, traffic market or others to do the base of Orning Oredas, because otherwise they can't buy our products.
Jeanette Nilsson:And that's also an interesting thing, because when you talk about, when you say language, when people say that data is gold and people, and then you don't really understand what the data is, they think, ok, they text this data, ah, I have a lot of text, I have a lot of gold Instead of understanding, ok, I want to do this process and then I need these kind of parameters and this information in every step, and I haven't ever collected that. I haven't, or maybe I have it, but it's not structured, so then it's not in use.
Jesper Fredriksson:But that's changing now with Genitive AI.
Jeanette Nilsson:Now.
Jesper Fredriksson:I think even if it's not structured, it's not a problem. You can still ask Chatchivity and put that in the context. So that is, I think, loosening up a little bit this problem. So I relate to what you're saying and what we've been talking about. When it comes to small and medium-sized companies, it's often I come in there as a data scientist and the first years there's not much to do because we don't have data. That's a problem. And then at some point, after a couple of years, then you can start to do AI because you have data. And I definitely recognize what you're saying. If you don't have it structured, it is hard to do something with it. But with.
Jesper Fredriksson:Chatchivity. I was here as a guest a few weeks ago when we talked about retrieval of augmented generation, which is the pattern that came out last year after Chatchivity so taking your unstructured data into a prompt in Chatchivity and then getting something back. So now I feel like it's loosening up a little bit and maybe it's easier for small and medium-sized companies who are in that situation at least to work with data.
Jeanette Nilsson:Maybe, Maybe Something that I often meet is that people have not kind of nailed the problem. What is the problem with my company? Is it that my workers don't do enough work, or the competition? Is too hard or we need more people. It's easy to say that we have to be more effective, but in what kind of where does it?
Jesper Fredriksson:hurt most. So you're saying that it's some sort of basic education on understanding data and the value of data and the value of data in solving problems? Is that a problem more?
Jeanette Nilsson:Well, I'm thinking like also what vision do we have for Sweden?
Jeanette Nilsson:Yes, that's a good one it's so easy to talk about. We are going to use AI for this and we're going to have For what For what. But I mean, what kind of country do we want to live in in 50 years time? And if we want to do that, then we know that we have different kind of tools, or what we will call them, to make this happening. And if we know what we are aiming for, it's the same in everything. If we know that this is the society that we want, we want everyone to be nice to each other. No more bullying in school, no more abuse, no more problems this is what we're aiming for. How can we do that?
Henrik Göthberg:But have we defined that? No, Because I think this is the core topic. You know, it becomes more and more clear to me. I've been following Mikil Dallian. He starts a new professorship in Handelsugskolan about well-being and happiness, so it's a very interesting topic. What is the AI utility function? What is the objective with AI that doesn't leave us to become the wolly fat guys miserable but actually makes the most for humanity, makes the most for us as a productive nation where we are happier.
Henrik Göthberg:I think this is very interesting because what is the objective we want with AI then? Is that answered? I'm not sure.
Jeanette Nilsson:I don't think it's that at all, and the problem is, like some people, sometimes companies can phone me and say Jeanette, I need to do an AI, I got this amount of money from my boss and I need to do an AI. I need AI.
Henrik Göthberg:Yeah, that's the problem you're going to solve. What's objective?
Jeanette Nilsson:Yeah, and then it's a pilot, and that's the problem with all of these pilots. No, as I think I would like more people to think okay, what do I want to offer if I'm working in the elderly care, Not trying to see, can I squeeze out more time of this? And blah, blah, blah Instead. Okay, what kind of care do I want to be able to give to people that are old in 10, 15, 20 years? I want to do this. Okay, what do I have to do right now? Because we know that there are a lot of inefficacy and ways that we can be smarter, and that also I have also found out because I have a lot of people from the union part of AI agenda. People want to contribute, but you also have to feel that you are important, Because if you don't feel that you are important, I think that's something that is needed from everyone and in Sweden, basically, as we talked in the beginning, if you have a job, then you are important, and that's maybe something that we have to talk a little bit about.
Henrik Göthberg:But I get a really strong connection with what you're talking about. Now we say we want to be strong as an ecosystem to drive a common focus, and if we learn now from how the best tech giants is doing AI, they have a laser, sharp focus on the business problem or object that they have and then they go out and seek the technology they need. So we don't start from the technology. Google starts from how do I improve the experience of Google Maps? And then they realize early on with Google search oh, we need to invent MapReduce, we need to invent the file structures that can support our business problem. That happened in the Stone Ages, right?
Jesper Fredriksson:This is the Stone Age, 15 years ago.
Henrik Göthberg:But it's actually a laser, sharp focus on the business problem. It's like I'm going to go to Mars, then I need to solve technology problems, and here we now go the other way around and we are trying to rally ourselves around that. But if I ask a simple question, what problem are you solving? We have a really, really hard time defining it and we know that as AI engineers. One of the biggest problems, I think when I look at AI problem solving or making value in any large company I've been working with the problem framing is so fluffy so you cannot translate it to a data and analytical problem. I think they haven't really. Oh, we need AI to improve customer or sales. Come on. We want happy customers, come on. Can you break it down, please? What is the core question? That is the big challenge we want to solve? Isn't that part of this whole topic?
Henrik Göthberg:We need to get much more knowledgeable and much closer to the concrete stuff of how this works, not so much for the technology's sake, but in order to articulate our objectives. I think we're poor at this.
Jesper Fredriksson:What do you think? Maybe, I don't know. I started recently at Volvo cars and I was surprised when I got there that there's already a bunch of projects up and running and it seemed to be most of them are really focused on improving existing problems.
Henrik Göthberg:That is nice.
Jesper Fredriksson:That may be an exception. I'm not sure.
Henrik Göthberg:How do you think about the AI ecosystem? Because, if we want to create focus, what are we focusing on together? What will rally us? How should we frame that?
Jeanette Nilsson:I think that is something that I'm trying to work with a lot of other people for some years now. We used to call it we need this Apollo project.
Henrik Göthberg:I explained that story.
Jeanette Nilsson:Sweden wants to be best in something.
Henrik Göthberg:We need to put a bold goal out.
Jeanette Nilsson:I like it.
Jesper Fredriksson:Who doesn't want an Apollo project?
Henrik Göthberg:No, but you need to rally around something.
Jeanette Nilsson:If Sweden just wants to, we will manage and it will solve itself. In the end, then I feel maybe not.
Henrik Göthberg:What is that clear statement? What could that be?
Jeanette Nilsson:I have tried to push this to saying that we could be the best in energy efficiency.
Henrik Göthberg:I like that idea Close with Vatnefals.
Jeanette Nilsson:Thank you. The first thing I did when I was in RISE was working with the data center research in Luleå that started after Facebook chose Luleå to put out the first non-US site.
Henrik Göthberg:I was at Vatnefals when this happened. I worked with the guys that made the deal happen. Shout out to Rick.
Jeanette Nilsson:Now there are three. I think they have a kind of building permit for five different centers.
Henrik Göthberg:I need to tell the story. When this happened, rick, who was the guy stitching this together? He was invited down to Silicon Valley to talk about much more efficient data centers. I think it was something crazy, like Jeff Bezos standing on stage. There was a conference for the Silicon Elite and they think you know the shit. Then he was like you know what they are doing. You know what they are doing with Facebook up in Sweden with Vatnefals. This is so cool.
Henrik Göthberg:It's like Jeff Bezos talking to the Silicon Valley Elite about this deal in Luleå.
Jeanette Nilsson:It's interesting because then I was working with Tobi Minden and others that had started the research. Everywhere I came, we went to Monaco and when I was saying I was from Luleå, everyone in the data center knew where Luleå was.
Henrik Göthberg:This was the first like. This was like are you saying you are using the code? This is so smart. It's more efficient and it's so green.
Jeanette Nilsson:Actually, that's something that is now in the I'm also working in. One of my European missions is to be an advisor to the European Commission on where to put out supercomputers.
Henrik Göthberg:This is another rabbit hole. We should go into the supercomputer topic. Anders, just prepare us for this. Should we take it now?
Jeanette Nilsson:I can just say that in the kind of plan now it says that it should be placed in where it's called national code, not in the summer course.
Henrik Göthberg:But just to sort of wrap this up, are we good at making objectives and what we are moving into now? Because now you stated like one objective, one bold mission could be we should be the best energy efficiency, or something like that, because I remember the first as an example how we have improved. I think the first topic that came out in terms of how we want to frame AI in Sweden, I think, is very vague. It's like we should be. We cannot be the best in the AI tech, so we're going to be the best in using AI and this is like it's like. You know, this is not a strategy you cannot put, you cannot. This is a platitude.
Jesper Fredriksson:It's so obvious, right? We can't build a large language model.
Henrik Göthberg:You know it's also like so, as you said, to me, if you can't put not in front of it, it's not a strategy. Strategy is about choices. So if you say we're going to have the best, we're going to have the most happy customers, who doesn't want that in a way? Right, so you need to put it. So I like what you're saying, Like we're going to be the best in what efficiency?
Henrik Göthberg:Okay, interesting, good. Or I go, jan Bosch in Schalmisch, sweden, europe. We are great in technology and our software industry is about embedded systems. We don't build large we're not going to build the large cloud providers but we are damn good at embedded systems. You know to multiven all that, so maybe that's our thing. Now we're talking right. I think this vagueness is one part of the problem and someone needs to do like you did now say let's focus on efficiency, energy efficiency, and make that our thing.
Jeanette Nilsson:And then you can kind of other things can connect.
Henrik Göthberg:Yes, then you can start unpacking the beast.
Jeanette Nilsson:So but I'm thinking that the problem sometimes in Sweden is that we say we want a leader, because as soon as someone steps forward and say I can be the leader, then you had to kind of, but what?
Henrik Göthberg:is that? Because now we're almost moving over to the next topic in terms of how to work, how to make success happen in Europe and how does this?
Henrik Göthberg:Finland does it? Because we have this sort of very decentralized way of working. That is very nice in some ways, but at some point we all we can't really. We get into a fight where we almost like we talk shit about each other instead of finding our spot in the team. I don't know how to frame that, but there's something here that I think is super important to understand what we need to do differently that you are on to here.
Jeanette Nilsson:Yeah, and I'm thinking also that we really need each other in Sweden. Yes.
Jeanette Nilsson:And a sad thing that I'm thinking also is that people say that Facebook is in Luleå, in the northern part. Facebook is in Sweden. I think that everyone in Sweden should be proud of that. Facebook chose Sweden, and if we could think more of that, for example now, when things are moving a lot with a green energy in the northern part of Sweden that's in Sweden. I mean, it's only we that think that it's in some specific part of Sweden. Other people doesn't know. For them it's Sweden, and then why do we?
Henrik Göthberg:do that? What is that all about? Is it the Stockholm versus rural versus Gothenburg? What is this happening right now? Why are we doing like? Is it the geographic is? We are such a whole country, so in the sense, we're we're, we're half the Europe in one country, right, is it? Why are we doing that? Because I really sense that you're on to something here.
Jeanette Nilsson:I think we do it because we can afford to do it, since we are thinking that we are such a rich country we can. We don't have to be. I mean, like we have 290 municipalities, some of them are small, like 2000 people, and they are supposed to give the same kind of services as Stockholm's, of course, of course. And people that are living in these small communities, municipalities, they want to have the same service because they pay the taxes they pay even higher taxes.
Henrik Göthberg:Yeah, so from a fairness point of view, they should have the same service, but we have. This is. This is something. Something is not really adding up here.
Jeanette Nilsson:No, and I'm thinking that we kind of have to be more to feel that we we want to help each other. We have we are in the same country, so in, because I've actually been in kind of discussions where people saying, if it can't be in my city, let's, let's the finn have it. Instead of thinking well let's go.
Jeanette Nilsson:Let's go to Hamburg have it, or let's Stockholm or Lule have it Instead of. That was something that I learned when I was working in Brussels as a lobbyist, because I was working for the northern part of Sweden and I was in and it was in the time when we actually changed the unslutely for drug to actually include the sparsely populated kind of we in Sweden we have more money because we have sparsely populated areas, and we did that together, we with the fins and with people from Spain. But let's park this.
Henrik Göthberg:Kirill, let's do the AI news now, because I think now your anecdote around these topics and how this works in Brussels is telling us something of how we need to organize ourselves, organize ourselves better in the AI ecosystem, because I think the learnings of what you learned and how it works and how the fins work different I mean, like that whole story you told me upstairs is beautiful. But let's do AI news first. And then we go there.
Henrik Göthberg:There we go, there we go. Ai news. It's time for AI news, so, shannet we have this AIW podcast. We have this section AI news. It's really started with the chat chip. You know the LLM craze. There was so much happening every week so we said, ok, what's the latest news that we want to sort of highlight that happened since last week. I know you have an interesting topic to talk about, I think. Are you ready to drop a scope?
Jeanette Nilsson:You're not going to get in trouble. Hope not, no one knows.
Henrik Göthberg:OK. So I think, don't you start with the headline news, so you start with the scope. I'm excited what I'm excited, so now I talked it up.
Jeanette Nilsson:But there's a short background before I say what is An interesting thing is that in the European Commission it's usually that you kind of negotiate with the member state first, but now, since chat GPT came, they feel like they want to be on to the next thing before anyone else says so they are kind of arranging meetings with people that they think have expert knowledge in the area to find out what is the next big thing. So we don't miss it in Europe.
Henrik Göthberg:So that's a different way of working here or thinking.
Jeanette Nilsson:Yeah, it's totally different, and but then if they can be agile enough to catch it. But so this Monday afternoon we had an informal meeting with people that are experts in the area of large language models collapse.
Henrik Göthberg:Yes, the big news. Yeah, so there's two news in one. So it starts with the news that we could highlight about the large language collapse. Could you elaborate on that? So now you have two news this was the big news and then we take it into EU. Can you do that?
Jeanette Nilsson:Yeah, and the thing that are now the experts talk told us these people well, maybe 20 people from the European Commission there were, showing that how large language modeling, the evolution of all of these kind of synthetic data that are poisoning these models, and how, if we are not able to have better input, better data, than these models are trained on, like the open models, and so on.
Jeanette Nilsson:So the highlighted collapse has something to do with synthetic data as a way to find more data to train on and I'm not super expert as you are, but the interesting thing is that we are now going to continue, next, have more meetings on this and going to have inputs to what can we do to avoid this collapse, to see that we have data that actually gives us models that we can rely on and that we can use it. Because if a different kind of expert shows also the poisoning that is happening with all this kind of fake information, synthetic data and even though it might be a good idea to try to, since most data are on, especially if you have health data, it's only on white male in a certain age, but if you are not, so it's easier, you have to. So it was really an interesting thing and they were really interesting in seeing how these kind of so really, fast.
Henrik Göthberg:So the collapse was just a couple of days back and they got experts and the expert was talking with the key people in the commission and now we are already sort of working on. Do you have a sort of a working group starting to slowly shape up to have more meetings?
Jesper Fredriksson:on these topics. Just pause a little bit before you lost me. What is the large language model collapse?
Jeanette Nilsson:Is that the collapse is when the data is kind of when you have so much data that has been generated and you don't know if it's true or not. So, when you're doing the next kind of model, you have a lot of data that you don't really know what it is, and then you are building the model on that.
Jesper Fredriksson:Some kind of vicious circle?
Henrik Göthberg:Was it in relation to when we used the word collapse? Was it referred to what happened with Gemini or what was the no? No?
Jeanette Nilsson:no, no, no, no, no. But that showed what it can do.
Henrik Göthberg:So this was actually decided before the collapse of the, when chat GPT went mad and also it was before that the core topic was when the chat GPT went sort of a little bit berserk.
Jeanette Nilsson:We had actually we did the kind of agenda before all this happened. So when this happened, it was like okay, we should have done this first, but earlier.
Henrik Göthberg:So you actually had the call. As this is actually in the car, that to happen. It's like so, so what is? How can we make the models collapse? So if you do this, if you use too much synthetic, that in the wrong way, it this book. This was sort of.
Jeanette Nilsson:Yeah, and the people, different experts also from different countries, showed the the problem when we really want these models to work yeah, everyone that works with models want them to work and want to have data that are high quality data that they can build these models on. So the result that we actually because in Europe we want to have fair, ethical, we say data models I not bias, I mean so if we kind of trust models and you started to talking about that are we going to build in our models, in the American models, or how are we going to do that? So? So this is something and something that I feel a lot when I work in different kind of contexts in in I have both super computers and and where I work in Europe we really want to try to find the European way.
Henrik Göthberg:Yeah, but, and to back up into what is the scope here, this is actually what. What happens now? When you are inviting people to search for the next big thing and talk about this, we actually looking at a slightly different way of working in the commission to start inviting to to be ahead of the game. Yeah that is sort of. So this is sort of the what happened here and what was done last week last week, so to speak. Monday this week. Monday this week. That is not the way it used to be done.
Jeanette Nilsson:No, and this is a great opportunity for Sweden.
Henrik Göthberg:Oh, let's talk now. This is the real scope.
Jeanette Nilsson:Yeah, I mean, like today, all people in a lot of people in the European Commission thinks that Sweden is the best country for large language modeling. Why, mm? Hmm, yeah, we have great people. We have people that we have done. You have been technology.
Henrik Göthberg:Yeah.
Jeanette Nilsson:Magnus Agri and Daniel Elblad and Freddy Kline, so we have a lot of people.
Henrik Göthberg:They know. Yeah, they've been part of the met him in the area. Yeah, they know them.
Jeanette Nilsson:They are well connected and are super respected. So, sweden, I get a lot of nudges that you should take lead in this Sweden, and so Sweden has the opportunity to take this lead. In working with large like us, I mean, we have also people have really high thoughts of our morality and the way they were not leaving people behind in the society. They were trying to have a lot of this yeah we have been working with this a long, long, long time, so we have a big kind of they have a credit.
Henrik Göthberg:Yeah, we have a credit. Yeah, super.
Jeanette Nilsson:So we. So that's why I think we could take lead in energy efficiency, large language modeling and also in this edge computing. And so I mean we have been Edge for the machine.
Henrik Göthberg:I know the guys at scale out, so we know these guys.
Jeanette Nilsson:So I'm thinking there are different, and so we have so much good things that we could take lead in.
Henrik Göthberg:But then we need Okay, let's park that now in the news. So let's come back to how we could take lead on this. I think that's a brilliant topic. I'm going to leave it to you. Thank you for that first news. What's your news?
Jesper Fredriksson:I actually have two different news that I happen to see. One of them is I guess everybody has heard about Sora by now. So the wonderful videos that came out of the open AI model and what blew everybody's minds I think yesterday or something like that was the one of the cases of this new model that's coming out of Alibaba, called the model is called emo. One of the cases is the, the Sora woman.
Henrik Göthberg:Yeah, it took the corner from Sora.
Jesper Fredriksson:It's called the Sora woman, which is from the I think it's from a Tokyo street scene with very special lightning, and there's a stylish woman with leather jacket and glasses, and now she's also singing dual dualy pass.
Jesper Fredriksson:So they took it one step further and it's a wonderful from a still image you can animate the face and the lip sync to the music. So it takes in the music and it takes in one still image and you can generate something that's moving. Is it possible to show it? To show it live? It is worth seeing if you haven't seen it.
Henrik Göthberg:Can you get some sound?
Jesper Fredriksson:So if you've seen this image, this video without sound, this is amazing.
Henrik Göthberg:That is.
Jeanette Nilsson:That's the next problem what to believe.
Henrik Göthberg:Because all this is super fake.
Jesper Fredriksson:Oh, fake, I mean it was AI from the beginning and now it's Now it goes. The AI generated still image of a video generated by AI is now going into the next model. So it's like you go from model to model and that seems to be like a common theme now that the models are becoming more multimodal and they take input from other models and there's cross pollinations everywhere. Oh no, wow.
Henrik Göthberg:AI generated. Is it any a or the Dualipa as well?
Jesper Fredriksson:like the sound is the sound, is the actual voice.
Henrik Göthberg:Yeah, and then they do the lipstick.
Jesper Fredriksson:But it's not just a lipstick, but it's like the whole face moves. Yeah, it's amazing. So that's one thing. And then they have. There was another model released from Google DeepMind, which was the Genie. That's a way to take a still image and generate your own interactive game out of it. Basically, so you can, you can model. The model is about interactions. So if you take that, maybe you can take the image of the woman and maybe you can, you can control her in an action game or something. So that's, that's one way of this, that's one direction. I see now that there's a lot of models are going multimodal and they're they're using something from another model as input. So it's super, super interesting to see where that will lead us, and I think there's also robotics going into this. So you're generating movement, which has the potential of the.
Henrik Göthberg:You can use spin and spin, spin, spin on this topic.
Jesper Fredriksson:Yeah, there's. There's a lot of implications, but I'm going to leave it there because I have another thing that I wanted to highlight that I saw yesterday also that was super cool and and even more techie and nerdy they. There was an invention recently which this build song called one bit, transformers. So there's a big thing about the size of models. In the large language, models are large, but one way to alleviate that problem is to make the unit, each, each number in the model. You can make it smaller, so you can make the it's as small as you can get, which is basically one bit.
Henrik Göthberg:So one bit transformer means OK. Could you say it in another way? So you're shrinking something else now.
Jesper Fredriksson:So, instead of usually, what you end up with in AI is that you have matrix multiplication. You have you have a bunch of numbers that you structure in a matrix and you multiply matrices together, and that's that's the basis of everything that has to do with GPUs you use are super good at this, but what's happening now is that we're finding a clever way to scale out the common factors out of the after these matrices and keep as small as possible the, the actual matrices. So, if you take it to the extreme, you can say that we only store not not just one image and not just one bit, but you have minus one, plus one and zero, so you have three values inside.
Henrik Göthberg:Three values in one bit.
Jesper Fredriksson:So they called one point fifty, one point fifty eight bits, because it's the log of three.
Henrik Göthberg:So what you're doing is now, you're doing the work, but you have managed to read, to compress or whatever you managed to compress.
Jesper Fredriksson:Which means that it's faster and it's also takes less space in the in the hardware and that's similar to what we've seen before with GROC, as in, not the Elon Musk GROC, yeah, but the but the LPU Exactly.
Henrik Göthberg:So this is an LPU.
Jesper Fredriksson:It's a specialized hardware to to be faster and more efficient in handling this. So this is it's not the same trajectory. This has not yet this one point. Fifty eight bit transformer is not yet specialized hardware. I wouldn't be surprised if it becomes, because this is a really big thing. We're scaling down everything and we're more efficient with this approach. And the the cool thing this if you, if you want to go really nerdy into this the cool thing that I talked about with one zero and minus one is that when we multiply that with another matrix, then everything is just. Instead of a matrix multiplication, that becomes just addition, which is a whole different level of math. It's much simpler to do, so everything becomes much faster and more efficient.
Henrik Göthberg:So what we're seeing here is that as we go more smart, like so, the way we accelerate is not used in the size, but it's in the techniques of compression, exactly the techniques. So when we say we're going to have more slow, like we talked about before in transistor and the way we've been measuring, sort of the more slow, of the logic of scaling large language models, now from now it's a clear trend. That is not that simple, because we are scaling the efficient. You know the index is a combination of the I mean now you are, you're, you're increasing the Capacity or acceleration in several dimensions or vectors is several sources. So in one hand side, so it's a combined efficiency we need to look at how big is the ball model, how compressed is it and how and the complexity of the so the three variables.
Henrik Göthberg:that actually in the end Exponentially probably accelerates the efficiency here. It's the typical you have. You have three different dimensions. This is in a normal compounding effects. This is hardcore compounding effects, like if I can increase that with 50% and this one with 50% and this one with 50%, the total increase is quite big.
Jesper Fredriksson:And the most immediate effect is that the models become smaller, which is more of a linear scale, but still it's very important. And when it comes to the, the actual calculations that you go from, instead of matrix multiplication, which is if it's the fastest version, is or do of log n times n.
Henrik Göthberg:Yeah, which is less?
Jesper Fredriksson:than less than n squared. But now you're going to something that's linear instead, because so that's, that's a half order magnitude.
Henrik Göthberg:The half order magnitude less, which basically then you know, because we were thinking how we have a problem now when the model gets bigger and bigger and bigger, and now we're solving that problem in another way.
Jesper Fredriksson:Yeah, and I can if I try to extrapolate a little bit from this. It feels like so we're seeing new hardware probably. Yeah. Lps is already there, and this is something new that might come out as a new hardware.
Henrik Göthberg:And we see a hardware optimized now for this architecture, all right.
Jesper Fredriksson:And if we, if we, if we draw it out a little bit, if I try to take Andre Carpathian vision about LLMs, he's seeing LLMs as the kernel of a new operating system. So we're going to end up maybe with a new computing paradigm where, where the computers that we have today is LLM oriented. It's a kernel just as well as the. The Macbook that that I'm using today is now in a different. It's using ARM instead of Intel. Maybe we're going to end up with a new computer architecture a new architecture for the LLM era could be.
Jesper Fredriksson:That's, that's really drawing this thing.
Henrik Göthberg:And now it's interesting now because I saw we're going down a rabbit hole here. But I saw, I saw. I saw Ted talk, or actually I saw a panel with Jack from Nvidia and he basically said we needed to make our bets on the architecture 10 years ahead. It sounded ridiculous in his space and this is now why right, how can you make a prediction on architecture?
Jesper Fredriksson:It's super interesting, for if you take the Nvidia perspective now, they're on top of the world right now.
Henrik Göthberg:Right now in this and we bet it on Betamax. So what, yeah exactly? Betamax or VHS problem?
Jesper Fredriksson:right. But the problem is, I guess, if I were, if I were Janssen, I would think like, how do I stay ahead of the curve? Because with the LPUs that's already there.
Henrik Göthberg:And it's probably has so much capacity in relation to this field, so it's easier for him. The barriers to entry to switch is lower for him. So to go from scratch, I would argue.
Jesper Fredriksson:Yeah, but it's, I think it's it's problem. It is a luxury problem, but it's still a problem if you're, if you're on top of the world and you're making something that everybody wants and you're saying that maybe the industry yeah, this is the codec moment.
Henrik Göthberg:Yeah, exactly, it's super interesting. Good, I have a couple of that's a great rabbit hole, by the way. A couple of more news we have. Have you all seen the Klarna? It came out so big this time in Sweden. The Klarna made out made some a lot of stories about how much efficiency they got out there. Ai chatbot have you seen that.
Henrik Göthberg:I think that came out one or two days ago and they're talking about sometimes that in the first two or three months in production it literally has done the job of several hundred people. So it's mind blowing stats, Absolutely mind blowing stats. Klarna was, you know, the financial provider in Sweden which are very, very tech savvy and they've gone all in and done this and they put it in production. And I've seen a couple of different. I've been trying to following the what's the social media commenting on this Interesting, yeah, and there's a couple of things that are coming out that comes in. I mean, like, first of all, some one crowd says you know what? That is happy thinking, you know it's too good to be true. Yada, yada, yada. One story of this. Then someone said another story which I think is quite interesting Klarna has made a point where they actually they don't allow the AI bot to go so far.
Henrik Göthberg:So as soon as the AI bot gets sort of used a little bit wishy washy, immediately it switches to human interaction and I think the general consensus on social media is that that is actually a very prudent way to do it. So you have a chatbot and you don't try to make it perfect, you don't even try to go for 60, 70 percent, you go for 40 percent, you go for 30 percent. Boom, it's 700 calls, blah, blah, blah. So they actually make it can probably do up to 80 percent accuracy, but they actually switch it at 50. So the way they've done that is prudent.
Jesper Fredriksson:How did they make the switch? How do they know when they should switch out the bot?
Henrik Göthberg:I don't know the details, but it seemed like. It seemed like. The way I can interpret from the text is that the algorithm is made in such a way, that is, if it feels certain of the answer, if it's a straight answer, if you can find the answer, it goes for it. If it doesn't have a certain answers, it leaves it, it immediately switches. So you know it's a void hallucination.
Henrik Göthberg:And then, of course, you had the third camp who basically said oh, you know what? We were sitting with this fucking robot. It was bullshit. And the best thing was we had to figure out how to fool it, to understand that it couldn't help us. We got to talk to a real person. So you know. So what to believe, right? So mega big, impressive press release, some interesting comments of how they are, you know, switching over to humans. And then you know one camp to say you know what, I've tried it. You know, the experience is not that great Bottom line, even if you scale down just a little bit, you know, or a lot, on what they think they've done super big savings. You know what's your thoughts here?
Jesper Fredriksson:I'm working with the case just like this on the mobile cars, so it's super interesting. I posted this as something to look into, something to be, at least to take inspiration from, and what we're seeing today when we investigate the cases that we get from email chat. We don't have a voice called transcripts yet that's in the works. What we're seeing is that this in the emails, it's quite hard to first of all figure out what's happening because people tend to write very long letters and some things are not related and it's harder to answer. But when it comes to the chat, transcripts. When people are typing, they type shorter right.
Jesper Fredriksson:Yeah, that is shorter and it's more to the point that it seems like they have a specific problem that they want to solve.
Henrik Göthberg:This is more programmable, so to speak, with the letters.
Jesper Fredriksson:So that's something that we can see, that, with the given resources that we have, we could make that automated to some respect.
Henrik Göthberg:And to some degree, I'm not surprised if, in reality, a large language model with a good training data can be way more efficient. Because it's also the speed right, because it's immediate, you don't need to think and type, it just comes straight away. So I'm not really surprised that the velocity is that much higher.
Jesper Fredriksson:Yeah, I mean velocity is okay. I mean you still have to make a call to an API somewhere if you don't have your own language model in-house.
Henrik Göthberg:So what's the turnaround time on a response?
Jesper Fredriksson:It's relatively quick 10 seconds 20? No, it's, so we're not running this live right now. But it's more like one to two seconds.
Henrik Göthberg:So your expectation is like two seconds later. It's super fast.
Jesper Fredriksson:Which is way?
Henrik Göthberg:faster than if you had another human typing right. What is your views on the Klarinhönn?
Jeanette Nilsson:I think this is a good example and I think that most Travikverket has also done a really good example when they're doing the understanding paper, when they are recruited, and it's always the same. The process.
Jeanette Nilsson:They made this robot processor. So if you find these, I'm thinking also that a lot of people want to do so much complicated AI. It's like, oh, I want to have a human eye why they put away dishes. You have so much mundane shit, Instead of saying, okay, this is something that occurs every day and I have 10 people that are doing the same thing every day. If I can automate this these 10 people they can do something else.
Henrik Göthberg:That's how you get more happy workers. I think in some way.
Jeanette Nilsson:So I'm thinking it's a good example, and especially that he shows also that he can scale.
Henrik Göthberg:I mean these I think this is the key point with Klarinhönn. Klarinhönn is the difference between doing Pox and doing something in production, and to show something and have 1, 2, 3, 1. A couple of million calls this is the topic right To build an. Ai system that scales. This is impressive. The potential is huge but people will use talk about it and then Klarinhönn goes out and does it. This is what impresses me. I guess the scale.
Jeanette Nilsson:I hope that could inspire more people to start working on, because I'm thinking, instead of trying to do this super difficult thing, start to think okay, we want to go to Mars, what will we do first?
Henrik Göthberg:Elon Musk he needs to get out of orbit. He needs to get out of orbit. He needs to get out of orbit. That's it right.
Jeanette Nilsson:So I'm thinking I think that is a good example and I think that we really need to understand that if we are not fast on this, we will be subcontractors and then we will not. Firstly, the bias we will not be able to tell what kind of values this bias will be based on. We can't sell things because we are all in the production line. Someone from another country is selling tourist productions or elderly carers.
Henrik Göthberg:Should we leave the news or do we have? I had a couple of more news, but we could also leave them, because we talked quite a long time about these news.
Henrik Göthberg:It's up to you. I want to share one more news, which is a little bit different but I think it's good, relating to our conversation here today with Janett. Did you see? There was a post from the Singapore Parliament? They are coming out with a new subsidy in Singapore. This was in the parliament on 26th of February in their parliament, and this went viral.
Henrik Göthberg:There is like a video from the parliament, someone speaking on AI and talking about the subsidy, and there are two things that blows my mind. You need to watch this. We'll take it later. First of all, how eloquent the person, the politician, is talking about data and AI in a way that is not platitudes, it's to the point, simple language and beautifully concrete, and I was like, wow, that's what we need People who are educated enough in their job as politicians to really understand what this is all about. That was the first thing when you watch this video. Then it was the subsidy itself, which I find super interesting. So on the 26th of February, singapore proclaims that they will offer every Singaporean over the age of 40 to go out and get a second, a new degree at the age of 40. And their argument is simple. It's a little bit like we need to really step up into lifelong learning and we need to step up that basically, what we learned in 20 years is not applicable anymore. Definitely.
Henrik Göthberg:How can we have our whole society work then? And then he made I got goosebumps too, thinking about it, because then he talked about what happens when it's bringing people of the age of 40 into university. Again. It's paid by the state, you're supposed to do it and you're then collaborating 18 year olds together with 40 year olds. There are so many spin off positive effects on this in terms of learning. So they bring experience into the university and the Yankee kids is bringing the mindset and the points of views to. You know, it's just connecting the whole fabric of society back again. So I think lifelong learning is go nuts on education and then basically recognizing that we really need to think about the whole. We need to think about the whole story here. It's not about redoing the education for kids, it's about lifelong learning now. So we need to redo the whole thing. And it's a bold move Everybody get a second university degree for free. Wow, isn't that cool yeah.
Jeanette Nilsson:Canada did that some years ago for women.
Henrik Göthberg:Yeah, but but and we talked with Eric Hugo on this podcast about this last week, where we think this is one of the biggest he saw this as shame on Sweden. He lived here for 27 years. He thinks Sweden is great. How the fuck can we leave our educational values behind free education for all the? We talked even on on the educational reform in the 1800s that really built the industrial. You know Swedish wonders. It built Nobel, it built Elamirix on it built all the big Swedish. It built the Swedish wonder. The 50th year of prosperity was built on the talent coming out of education reform that took us out of poverty and you know farming, to be honest, and the core game was not the elite but it was education of the masses.
Jesper Fredriksson:I think. I think it's a great thing and I think also that it's a little bit up to the individuals. I think we will have to as individuals be, lifelong learners yeah, we need to.
Henrik Göthberg:we need to switch on. I mean, like the state can do so much but you lead a horse to water, but we need to drink water ourselves. What do you think about all this?
Jeanette Nilsson:I think it's great because when I was working with gaming, I got connection with Canada because they also paid people to pay for the food. They also paid people $50,000 to come to Canada and start their gaming company.
Henrik Göthberg:They went nuts right and they have the AI wonder today.
Jeanette Nilsson:Yeah, and so they also saw that there was an inequality in the tech world. So they started to do three things. One was to have more tech for girls in kindergarten, trying to stimulate the interest. And then in the school, when girls are about 14, they saw that even if they were really good at mathematics and logic, the kind of just didn't continue because something happened. And then they so they focused on that and then they also gave women the possibility to go two years in education to learn more. If I was a nurse then I could have go and learn more about that. So when I came back as a nurse I could use all these because the skills that I have from technology and computer combined with the domain.
Henrik Göthberg:But when I see some of these topics, they are so obvious and there are so no regrets. So do we really need to think so hard? Is this something we need to have a political debate across the blocks, or is this so obvious that some of these things we should use focus on getting down? What do you think?
Jeanette Nilsson:Well, I'm thinking what you said every person is really can actually do this.
Henrik Göthberg:Yes, that's true.
Jeanette Nilsson:And also I'm thinking that we could either think that this is my responsibility to be well, have a good life and be interesting in the labor market, to see the sure that I actually and it's like not knowing English when I'm talking I have a lot of groups with younger girls because I'm really interesting to see that this new kind of game changer. So few women are part of this and it's even this trend in education less and less girls are choosing to starting on technological education and we see that in the future people need to have much more competence in tech. So if half of the population decides that this is not for me, then we have an education.
Henrik Göthberg:And then really, we don't know. It's like why don't you have more women on the pod? Why don't you have more women on the innovation summit? It's really really hard, both, you know they're not enough, and then in the end they don't want to do it, they don't want to speak on stage. Yeah, that's crazy, really crazy stuff. So we get a lot of flogging around these kind of stuff. Oh, you need to balance this up. Yeah, we want to, but you don't want to be here. What's wrong with you? You know so. So I think this core topic how do we flip it? So the women or girls want to have tech as the main?
Jeanette Nilsson:Yeah, and this is the tricky one- and you have, also because I know that, since I'm send, if someone asked me, do you want to be on stage, I always say yes, but you are different.
Henrik Göthberg:You can get in the sense I think you are.
Jeanette Nilsson:Yeah, and I want to be a role model for others and I say to people well, you know, it's what's the worst thing that could happen. Yeah, exactly, they will never kill me. Sometimes I feel like I've done something super great and sometimes it's not so good, but if you never do it, then it's like I. And anecdote also about since I have I'm a mother to boys and I was working at the National Road Administration and the we had a really good, well, the highest, the top dog. He was really a good guy, but he had accidentally said that it. You know, a lot of young women left the office and he said, ah, that's a shame, but we have all these great old men. So it, he, he didn't mean that, but it came out wrong.
Jeanette Nilsson:So we have this kind of he put out a group and they also oh, we are now going to work with equality and can't you lead this? And then I had done this before and I felt I don't want to be no, no, I didn't want to be part of it, because I felt like oh, and they said, oh, please come, come, come. So I came at the third meeting and when I came into the room I was a bit late, and then my, my boys. They were two and five years old, or now it was younger, one and four, because then I was in Brussels, but it won't fire, okay, so, and so they were really young, my boys, and everyone, because I have a habit of having kids and I, so everyone knew my, my boys, and it was this kind of feeling in the room that all they were only women and it's all all these men. They are never letting us and it was like they were really a lot of hate, frustration, frustration.
Jeanette Nilsson:I was I? I, I called it even hate. I was like, oh, there was so bad, it was like they were evil on the world. And then then I said then I just stood up and they said, oh, are you leaving? Yeah, yeah, I'm going home and I'm going to kill Oscar and Gustav because you say all men are evil. And then I think all, all boys should be killed. And they're like what are you saying? And I said, well, if we say that all men are evil and they are all, so that's the no, no, no, no, yes. Then I thought, okay, what have you ever done for me? I'm also an engineer. You have never helped me. You're always helping the guys. You have never helped me. So we got to start a discussion. What are we doing ourselves Instead?
Jeanette Nilsson:of saying the men is never going to let us out. I said what have you done?
Henrik Göthberg:lately.
Jeanette Nilsson:We have to take space ourselves.
Henrik Göthberg:And then, of course, this is the thing we Like you can give space, but you also need to take space. Yeah, yeah, it's a two-way street, yeah. Interesting, yeah, and also that's tricky to talk about sometimes, right. Yeah, because you're not politically correct when you say that. No, I know.
Jeanette Nilsson:But that's something that I'm not. But I'm also like to provoke discussions because I'm thinking that every person Now we have two provocative statements, at least today.
Henrik Göthberg:Yeah, yeah, yeah.
Jeanette Nilsson:Oh, I have so many more. Can we have one more, one more?
Henrik Göthberg:This is fantastic, because this is really interesting. What is your take?
Jesper Fredriksson:Yeah, I'm thinking You're mentioning that you want to be on stage. Was it always like that, or was it something that changed you? And what do you say to your female colleagues if they doubt should I be on stage?
Jeanette Nilsson:Yeah, I'm often saying that because I perform as a ballet dancer, so that's why I'm not so afraid of stage. But of course that's actually a good thing, right.
Henrik Göthberg:That's something that you had really early in life. That probably shaped you a bit.
Jeanette Nilsson:Yeah, I definitely think so. But I'm thinking I have a lot of adepts talk about this because I'm saying, of course you're nervous. I mean, think of what you want to rehearse, stand in front of a mirror when your heart is kind of feeling like it's jumping out of your chest, think of doing some counting. So there are a lot of different techniques and everyone has to find their own.
Henrik Göthberg:But in the end, you need to put yourself out there, because will you fold to the pressure, will you be scared, or you want to step up and face your fears, whatever they are, and I think that's the big thing here.
Jeanette Nilsson:And then I'm also thinking of since when I was in this kind of super male environment, when I started to work and I once I have so many anecdotes from this one all crazy. But one of the things that I realized after a while was that I could do a lot of the fun things and a lot of the hard work. Where the guys were the kind of we were recruited at the same time and after a while it felt like they were really harsh on them. They were on the boys, on the men.
Henrik Göthberg:They got the tougher.
Jeanette Nilsson:They were always so nice with me, all the men, but they were really on these crown princesses. They were really hard on them. They were kind of seeing if they could.
Henrik Göthberg:If they would crack or not, crack or not.
Jeanette Nilsson:And that's something that I'm telling also to women that that's part of the game. I mean, you can't be.
Henrik Göthberg:You're not going to get the top job if you haven't been tried.
Jeanette Nilsson:No, and that's I mean. It's hard, it's even hard, so it's. Sometimes I feel like some women think that you're just a man and then everything is kind of served. You just go into a room and then you get the executive. But it's not true. Men has to work as well, but of course they are in the norm. So it's a bit easier, but it's not for free.
Jesper Fredriksson:But I'm thinking again about sort of the same question about this going on stage. How do you handle that if you have employees, for example, that feel like this is hard? People are being mean to me. How do you motivate them? How do you tell them that this is part of the game, Because it sounds a little bit like hard? Can you make it into something that they can understand and work with? You see where I'm going.
Jeanette Nilsson:Yeah, definitely, and I'm thinking that's why it's so good to try to have some mentors and understand that, okay, this is If I want to reach my goals. If I want to do this, there are different techniques, and how do I maneuver in this? And when these different backlashes, like someone says to me oh it must be so fun to have this interesting job that you have, Shannet, but what about your boys? Are you seeing?
Henrik Göthberg:them enough. So you get that female question back at you again, right?
Jesper Fredriksson:That's funny, I never get that question, that is still interesting, right, because that is saying something.
Henrik Göthberg:You get that from the guys or the females?
Jeanette Nilsson:No, it's from the women, of course.
Henrik Göthberg:So women want to lift you up, but they still want to drag you down into the same topic. It's not interesting they should really be ashamed to ask that question. If they want to go after the same gamers, they should ask that question then. They shouldn't put that on you more than their husband, so to speak.
Jeanette Nilsson:No, but the interesting thing is that in this society women are also bearer of norms, have to really understand. Okay, for example, when I've been traveling a lot. This is so true.
Jeanette Nilsson:My husband. He has. Well, sometimes when I phone home, I remember once my oldest son, oscar. He was going to do something in school and outflicked. Then he said, yeah, I didn't have anything at home when he was going out in the morning. So he had two kind of hard bread macarons with him and smör and I felt like, okay, well, that was good. And I felt like, ah, why didn't you go out? But I felt like no, if I now go and say, no, I think you did wrong. Why didn't you go out in the evening and fix something? Then I kind of Then you have he sold it.
Jeanette Nilsson:It was not perfect, but if I had said to Oscar oh, you have a really bad father, he didn't go out Then I would not.
Henrik Göthberg:You actually programming yourself back into the same mother role again.
Jeanette Nilsson:I think he's great, he and I mean.
Henrik Göthberg:Can I get you to talk to my wife about these topics? Yeah, because I will never. I fully recognize that of course my wife should expect the same standards from me as she can deliver on these topics, of course. And at the same time I'm not as good at this. I can feel sometimes on these topics I have no excuse. I have no excuse. But what happens when you get into what you highlighted now, the way you manage that even the poor man, he did a shitty job, but put them a little. I think that's very smart. I think you will then grow together.
Jeanette Nilsson:If I would have liked, like the first picture, he had, oscar had a shirt. Like I said, oh shit.
Henrik Göthberg:Yeah.
Jeanette Nilsson:I like his school picture and his dad.
Henrik Göthberg:you'd let him wear this.
Jeanette Nilsson:Yeah, I mean, and then I realized if I was starting to say something or putting out clothes, oscar will have these.
Jeanette Nilsson:Then you put yourself back in the box immediately, yeah, and if I want other, because my sons has, since I've been traveling so much when they were small also, and my husband, paris, living in Stockholm, sometimes took them with me in the airport to Stockholm and grandfather, grandmother, took them. And I remember once when I was in Brussels and I was phoning to my grandfather the boys' grandfather and I said, oh, what have we been doing today? Yeah, we have been in the forest. What have we been doing? Yeah, they've been using knives. I like Okay.
Jeanette Nilsson:Boys will be boys.
Henrik Göthberg:Grandfathers will be boys, yeah.
Jeanette Nilsson:And they didn't they didn't die they didn't die. So I felt like, okay, if I'm trying to kind of, everyone must be able to do that in their way and it's good.
Henrik Göthberg:But you had to. But I applaud you, shannet, for this thinking, and I think it's so hard. I think it's so hard, but I think this is the game to play, because if you play the game, that Because maybe you are better at it and I don't want to go into why you're better at it or why men should be better at it or should not be so bad at it but it's simply playing the game, because if you're playing the game the wrong way, actually you're putting Everybody's putting themselves back in the old same boxes again, and I think that's the refreshing thought that you are trying to push here is that I need to bite my tongue a couple of times or else I will put myself back in the box where I started on these topics. Is that a fair summary?
Jeanette Nilsson:Yeah and also thinking. I have a friend that he moved in with a new woman and he kind of arranged everything in the kitchen and then she came and said, no, you have done everything wrong. Yeah, exactly, and he felt like, well, I'm going to leave here. I was kind of compromised. No, it's going to be my way in the kitchen.
Henrik Göthberg:Yeah, okay Then obviously this is your kitchen, and then how can you go back out of that claim?
Jeanette Nilsson:No, no, no, you put yourself back in the box. But that's also an interesting thing is that we are talking about power, and when in the school they had this discussion when Oscar was little, that we had all the boys' mothers in one group and all the girls' mothers in one group and one of the questions was do you, as a woman, have power? And if you were agreeing, you kind of you would raise or you would sit down and remember, but everyone else did one thing and I did the other thing. I think everyone, if you don't have power, stand up, and everyone stood up except me and I said well, hello, we are all mothers to boys. Do you? Don't think that we have power?
Jeanette Nilsson:And if you think so, then I think that you really, really need to start discussing with yourself, because everything that you do, if you are a good role model for your boy, he will be a good man. But if you kind of, because I meet so many people that think they are doing things the right way instead of doing what they're feeling, and I have so many stories about that. So I'm thinking that sometimes women think that they don't have any power and in some areas like these are my kids. I don't want to let them man in. These are my things, instead of thinking we have different kind share all the power. I mean my husband does a lot of decisions that people would think that I would do, but I say why he has done all of these things. I think it's perfect. Good, it would not have been the same as me, but I love it.
Henrik Göthberg:Because I think that's sometimes the difference that I genuinely think I've done a good job and I think I sold it differently. And, depending on which yardstick you use or which, you know, what is the value you are trying to teach your boys. You know, you could argue that my way was better, or my voice, and I think that's maybe sometimes that we are looking at it as like oh, you didn't do it as I did it. I've done this for 10 years now and you did it differently, so you didn't live up to my standards. But I was like but I wasn't really going for your standards. I was actually going for if the boys don't fix their own lunch, they don't eat. That was my standard, but you didn't fix your lunch, yeah, but I show them, I discuss with them in the morning, that you need to fix your lunch now and we don't have so much at home, so you need to be creative, you know. So it's like, but when you try to explain that actually I was going for another objective here, I think that discussion becomes very tricky immediately because like oh, but you know what, when I do this, the lunch is perfect and this is like this and this is like this and then like yeah, but I kind of think you're curling the boys right now.
Henrik Göthberg:I don't really like what you're doing. You know, in all honesty, I'm not really going to play that game with my boys. And you want me to play that game because you think that's the yardstick. I have another benchmark here and then we have a real. Then this becomes really tricky, I think, because I have, I can never win this argument. You know, there is really tricky argument to have deep in my bone. I feel I was trying to do something else, but I can't. I can't articulate it. Do you see what I mean?
Jeanette Nilsson:Yeah, definitely, and the interesting thing is that I think that is a problem that a lot of women had to really face and in our family it was like I came home and my husband always says, yeah, I have decided now that the boys will do their own washing.
Henrik Göthberg:And he's being lazy, right, you think it, but he's not.
Jeanette Nilsson:No, no, yeah, yeah, exactly, I said great, great, yeah. I bought a blue box for Oscar and a green box for Gustav because we had these color differences with the warm Montessor kids, you know.
Henrik Göthberg:Good, good Good.
Jeanette Nilsson:It worked out perfectly. Yeah, yeah, yeah. So they really yeah, yeah, yeah. And since I didn't interfere, I had just said great, when the boys they did the washes. And then a couple of weeks later it all said, yeah, I'm thinking now that they should start to do a meal once a week. I like, yeah, great. So in one year we had Macaronne and Chatboula once when the little boy was doing it. I'm really picky with food and the all the brother here, the piti panna.
Henrik Göthberg:Monday and Wednesday. It was crazy.
Jeanette Nilsson:But after a while my husband then found this Lina's matkasse. So that was really great. They came with recipes and food and the younger boy he called it Lina's hot cuss because he thought it was so difficult. And I said well, it's good, you know, you're learning to read recipes and things like that.
Henrik Göthberg:So now I think that was good. May I ask how old the youngest was then?
Jeanette Nilsson:Yeah, Well, what kind of been was it? Maybe 10?
Henrik Göthberg:Oh, that's good. I'm really impressed with that. Yeah, that's impressive. That's impressive. This is way harder than AI.
Henrik Göthberg:But I don't know, we are really running out of time. Can we steal some more time? I love the conversation because I really want to go back and I want to hear your short version. How do you make decisions happen in European Union? What did you do? First you thought it was and then? How does it really work? And what can we learn from that and bring home to the Swedish AI community? Because the story we talked about earlier before the pod, I think was super interesting. It's really straightforward when you think about it, but we are doing it wrong. So the whole thing with contrasting of how we in Finland or how we go into this topic of putting our agenda forward and how decisions are really, really made- I think, not many people really think about how this really works, because it means we need to understand the game and play the game in order to have any influence.
Henrik Göthberg:So that's where I'm coming from here. So I think you have so much brilliant knowledge about this. You have proof in the pudding. You put some legislation through. You talked about what do you call it Sparcely populated? You get that in the EU for ordnance. That is amazing.
Jeanette Nilsson:That was the result, and if I take an example of something that people maybe know more about because this was back in the days more than 10 years ago, in Sweden we have only forest.
Henrik Göthberg:A lot of forest and in the EU they want to protect forests Because they have some kind of forest.
Jeanette Nilsson:Someone also asked how many forests do you have in Sweden? I like one. Sweden is a forest. All of a sudden people realize we are going to make a new decision that is going to be bad for Swedish and Finnish industry that are really using forest. And then you try to influence the European Parliament because now it has been in so many different, because there is a way of every decision, how it's made. It's a very detailed process. It's a very detailed process and you know actually, if you've been working with that, I will come back to that. But often you jump to the very, very end. But it's very, very, super hard to make any decisions, any changes in decisions. And if you are going to try to change a political decision, like you have in parliament, then you have to understand that you maybe have to talk to over 100 persons. You cannot talk to the Swedish guy and say, hey, don't you understand that if this is going, to.
Henrik Göthberg:You need to influence everybody. Pushing a button to somebody, we almost write.
Jeanette Nilsson:And no one will do that because you have different layers. So if something Now I'm just making something up to have a discussion, an example, it will be restriction on using water, it could happen. And so they are talking about it politically. And okay, that discussion. Maybe we need to restrict how much water someone can use. Okay, two liters a day.
Jeanette Nilsson:Some person in one of the directorates, like in a department in Sweden, the department gets the. You have to do the investigation, you have to do facts, you have to take in a lot of input and starting to do a kind of suggestion on how many liters of water shall every person in Europe have, and it looks like okay, how many liters do someone in Spain use? Okay, one liter. So they make a decision on a suggestion and that kind of then travels into this decision chain where you sometimes have people from the national ministries that are negotiating through the permanent representation. In every country you have different kind of experts and every. Now, this was such a simple thing, but in this chain of decision, different levels are kind of designing simple things.
Jeanette Nilsson:So if you, For example, if this is a big thing and the simple thing is how many liters of water? No one on the top level are going to talk about these liters of water because they are talking about regulations, Blah, blah, blah, blah blah. So if you have this information, in Sweden actually we need 10 liters of water because we have all these kind of differences that we need an exception. Everyone else has one liter. We need an exception.
Jeanette Nilsson:We don't have a water problem, but we really use a lot of water due to, yeah, and we have a lot of, then you can see, okay, when you then continue, sweden from the beginning has this Because we're based on facts and say, yeah, we cannot. Sweden has to have this kind of special treatment because of this and this. And this is not going to rub the kind of disturbed the whole Congress. No, so if you're starting to talk to persons that are actually doing the work, then, it's in the text and then it's much easier.
Jeanette Nilsson:And, for example, if, for countries like Finland, they are having Since this is both a kind of official you're working with member states and you have decisions that are in this change but you're also working with this different kind of informal decision-making it's like the meeting that we had on Monday and organizations. So these are. If they are also saying, for example, that Sweden needs 10 liters of water, it feels like yeah, it's important, Sweden will have these 10 liters of water.
Henrik Göthberg:But if I contrast or relate this to daily life, working as a consultant selling to big corporations, I have worked as a salesperson and I worked on decision-making Wattenfahl, I think what you're talking about is really bottom line, a couple of core things here. Step one you need to know the decision process inside and out. And then when you talk about a decision process, you need to understand what is the formal decision process steps and what is the informal decision-making stakeholders along that journey. So know your decision process. You need to know the facts also Step one. Step two get on to the people who are preparing the decision material.
Henrik Göthberg:So in this sense, the only way to really contribute is to basically make life easier for the person writing something or supporting them with the facts, or basically getting your faxing. You collect the facts, you have the facts in order to shift the buying criteria or to shift the criteria or to be part of setting the criteria. If you try to get in on the end, the criteria is set really hard. So and then, thirdly you said, is the difference between formal power and decision-making authority versus informal influence? And to navigate this? Now, if you don't understand these core things, to go to the guy writing the stuff, understand the process and actually who is going to influence what part of this chain? You have no clue what you're doing.
Jeanette Nilsson:It's not a fair summary yeah, and also, if you understand the chain and you see that, for example, we predict that we take this water thing again, we predict that it will be a schedule of water in and this isn't an asset that Sweden wants to preserve, this is something that we are really then we maybe shall start talking about how many liters Then you bring up the topic, you highlight the problem, you drive the problem question.
Henrik Göthberg:So this is really also about don't start in the end, start with. If you want to really drive something, you need to start way at the beginning feeding the problem, feeding the problem that leads to a proposition.
Jeanette Nilsson:Yeah, exactly, and that is something that we could do in the AI ecosystem in Sweden.
Jeanette Nilsson:If we, for example, if we take this energy efficiency I mean it's so many parameters.
Jeanette Nilsson:That is important about talking about this so if we would like to have that we could start working on now I know that our work done on energy efficiency, but so that would be something that we could do with the people that are ministers, and something that Finns also do is like, even if it's left or right, they have, if they are, for example, when they decided to build all of Finnish Lapland with I mean, I don't know if you have been there, but it's amazing. You are in nowhere and they are like to 20,000 beds in a city where 3,000 people lives, but they have this kind of log houses beautiful, it's so crazy impressive and they built that for 20 years. And in these municipalities, they have used money from the European Union to build infrastructure. They didn't kind of change every fourth year because they knew that we are going to build this kind of. This is the city, the vision that we started to talk about, and then it could be different ways towards this.
Henrik Göthberg:But it seems like they get over the political differences and they go for it. You know, they get over the regional differences, whatever, and they in the end stick to one plan, one story. And you said it before. Doesn't matter who you talk to, they will all tell the same core message.
Jeanette Nilsson:That is different and for them also it's like Finland. First, when we were working with this partially populated fedraget it was most of my colleagues, I can't take so much credit of that but when I was at different meetings I met people from Helsinki and said, oh great, but you are working for Helsinki, this is for Finland. They said. And I'm thinking that it's interesting that we kind of regionalize things that oh, this is good or bad for Skåne or good or bad for Northernland.
Henrik Göthberg:We do that.
Jeanette Nilsson:We do that Instead of thinking this is great, the green industry in Sweden great.
Henrik Göthberg:But how can we relate that back to AI? So how could we find the common theme and a common voice in AI then? What would that be and how? Would we may need to work differently in order to come to that key message and then stick to that across political and regional borders. What is the difference we need to make now, in your opinion?
Jeanette Nilsson:I think that actually I don't really know where the problem is, because everyone says this is not a political combat, everyone thinks this is good and some people said this is why nothing is happening.
Henrik Göthberg:Yeah, because there's no debate. We don't want, we don't want to lose. Put voters on it.
Jeanette Nilsson:So I still believe and now that we have this AI commission that I know they are going to try to be very concrete and do things. It's not just talking about things and doing new papers and investigations, because we have done it a lot. So I'm thinking it's more of how to get this to fly is to start using and also be prepared to, like in Klara Nafal, not trying to solve the hardest problem, Starting to say this is good enough.
Henrik Göthberg:This is the first problem, not the hardest problem. What is the first problem?
Jeanette Nilsson:How can we and then kind of level up to Swedish national level see, ok, we have all this infrastructure, we have municipalities, we have all of this. What can we start doing that gives the first best results?
Jesper Fredriksson:Sounds like a start-up mind, like do the simple things first.
Henrik Göthberg:Do the MVP and then the MVP, and then iterate, but OK, let me, let me. Okay, in this context, what would be your advice or vision? For we started up a new Swedish AI commission quite recently, led by Swanbury. We have several friends. Heinz is in there, who else is in?
Jeanette Nilsson:there, are you there, martin Svensson?
Henrik Göthberg:Are you part of the no? You're not, no no, no no, what do you hope? What is the best outcome of how to use the AI commission that is set now so it doesn't become another report that's going to say the same stuff that you kind of said already in 2019, with 25 topics, yeah, what is the maximum impact value that we could hope for that the AI commission could sort of change the game or gel us together. Maybe. What is your wish that could happen with the commission now?
Henrik Göthberg:For me it's a fear that it's just going to go one circle around the table, or is it? Oh what if they could do this? What would that be?
Jeanette Nilsson:Yeah, my hope would be that, instead of doing more reports, they would kind of I mean, what we need in Sweden? We need the Swedish authorities to start working, because then companies can learn. Because we don't One of the things we have to face people don't know what this is, people don't know how to use the tool, people don't know what they want. So we have to start educating and learning. And then we also have to start working with IT companies so that if I'm a big, if I'm an official in an authority, working together with the ET company, so that we can both learn, Because otherwise, if we are going to, we must, on one hand, protect the companies the small, medium-sized companies that we have, so they can learn and not be kind of we take things from others, so that the and also that it will be regaining spirit to every authority how to use and how to combine, because they are so get it all the way down into directory of each authority.
Henrik Göthberg:Regaining spirit? I don't know, I don't know what it is. So regaining spirit is like the core directive that the board gives the authority.
Jeanette Nilsson:What is their frame? The minister sets to the different. Every minister has different kind of authorities.
Henrik Göthberg:Yeah, and this authority is then it's a directive that highlights the regaining spirit and what? They should do and, basically, one key goal could be to gather clarity on AI inside in relation to their objective and authority.
Jeanette Nilsson:Yeah, and I know I've since already said that Patrick Ikemua has been here. They have also done a lot of assignments from the government to see how this would be, because some things almost everything that you do together is better, so there are a lot of ideas on how to do things together. I mean especially a lot of small municipalities. They can't do this and they may be the ones that need the results most.
Henrik Göthberg:But maybe that's how do you think about this core topic If we pick. I mean, it's like the same in a large corporation when you get started. It's very hard to put out a little bit of money on AI to 100 different bosses around the company.
Henrik Göthberg:And no one can do anything, because basically, you need to basically make an effort, a concerted effort, an orchestrated effort, to start building infrastructure, to start building a cohesive strategy, to start building a cohesive pattern. So how you build stuff, if we go back, I mean, what happens when 50 different, if everybody is doing their own thing? It's used at Hockercy and maybe that's one key topic now. I mean, like that, you know, we need to find a focal point. Who is leading the way or who is setting? How are we setting standards of this? Because we are basically point solutions all over. But is that part of the problem here, that you know that we need that? Someone needs to decide where do we central this focus? There's no center point. I think that's the problem, in my opinion.
Jeanette Nilsson:Do you agree, or do you see what I'm talking about? Yeah, yeah, and hopefully the AI commission can kind of point out, and I'm also thinking that in Sweden we have to be kind of then also generous. Okay, if you can decide on this, what my children will have tomorrow on cloud, that's okay with me. It's not what I would have preferred, but it's okay because we are moving forward.
Henrik Göthberg:But this is the core problem back in. You know, we want to be decentralized. But if we don't want to federate around decentralized approach, we don't want to have standards, we only have a bunch of anarchy point solutions. So the challenge is completely centralized monolith, we would say technology monolith or we need to have a federated platform architecture. You know what I mean. Right, you can apply architectural topics to these problems, I think.
Jesper Fredriksson:I don't mind a little bit of anarchy, though no but what happens is better than nothing.
Henrik Göthberg:Yeah, it's better than nothing, but can you get anywhere in all small part? Can you go? I mean, I think the challenge with agility at the extreme is that you have the coordination. Costs are simply massive, right, but the monolith is too slow.
Jesper Fredriksson:I'm trying to draw parallels to where I'm standing. I'm saying it's better to do something than to do nothing.
Henrik Göthberg:This is the whole bond.
Jesper Fredriksson:That's a start and of course, it's better if things can gradually become more cohesive and you find a common strategy. You don't start with it. We're standing now. I think it's much better to do something than to wait for better solutions you can salary administration.
Henrik Göthberg:Decide this is one.
Jeanette Nilsson:This authority will do all of this. There are things that are similar that everyone has to do so that we can decide okay.
Henrik Göthberg:We're not going to have 10 different salary approaches.
Jeanette Nilsson:Let's do one.
Henrik Göthberg:But our decision making is not set up that way.
Jeanette Nilsson:No, and I hope that we don't have to kind of go back to being poor before we start. Because why is Finland so keen on this? They started to see, if we are using AI, we are going to have this standard of. Why don't they have our?
Henrik Göthberg:problems? Why don't they have our problems like we do? Is there steering mandates of the different authorities different? Is there starting point very different, because I find sometimes that we don't have the directives and the way the structures are set up. In Sweden it's really really hard to do centralized efforts. Don't they have that party? Is it a different structure or is it simply a different mindset and the way, the what is different?
Jeanette Nilsson:They have been in wars. Sweden, that's a hard question. We don't want to copy that one. The second world war? I don't want to copy that one. Everyone knows that someone died then. They have had hardships in Finland.
Henrik Göthberg:So they have to take a CSU bringing them together on key topics.
Jeanette Nilsson:And they know that they have to fight for themselves.
Jesper Fredriksson:So we were complacent.
Jeanette Nilsson:I remember when I was working a lot with people from Finland, interesting Well, yeah, yeah, yeah you know you're big Swedes.
Jeanette Nilsson:Big Swedes. And it feels like, yeah, and I've been in meeting with the people around the Baltic Sea and I was with someone from the ministry he's not working there anymore but then I said, ok, so it's going to be in English. Then he looked at me and said, of course not, it's going to be Swedish. It was in Swedish, but very few people really understood. So it was. I mean, since those areas had one Sweden was big Sweden, it was Sweden and everything. So, yeah, and then I realized, ok, I understand now what people in Finland say when they say, oh, you are big Swedes.
Henrik Göthberg:They mean it. They're not cynical with us. They mean it because they remember a sort of yeah, but it was cynical.
Jeanette Nilsson:It's like, ok, you think that you're still like you were when you were.
Henrik Göthberg:Yeah, yeah, yeah.
Jeanette Nilsson:Yeah, instead of saying OK, this is happening, we really need to see how can we help industry best. I mean, I remember when I was in the national road administration and really were fighting for different kind of infrastructure to Sweden, and people in the ministry said now we think it's better, I mean, for them to have them instead of us in Sweden. And I felt like, yeah, but if we are the only country that thinks so, then is this going to be good. No, so I'm thinking that, I'm hoping a lot.
Henrik Göthberg:But you said before they at the end, when they decide on one topic, the other people rally up and they take this place in the queue. They basically, they see the greater good and they go with it and they are not sort of bickering about it.
Jeanette Nilsson:No, that's also the problem in Sweden, like we are kind of fighting its shudder in Sweden Instead of thinking OK, sweden is one country, I will everything that happens in Sweden.
Henrik Göthberg:So we're exorting our energy on the wrong? Yeah, I think so.
Jeanette Nilsson:And fat lazy cat, instead of thinking that we really this is a chance for us and for our children in the future Working together, we're getting on overtime now, so I think in one way that could be the ending.
Henrik Göthberg:It's a good ending. I'd say it's a good ending and you know, anders is going to kill us because we didn't get to the philosophical questions then. But it was a good ending there. I think we end here.
Jeanette Nilsson:Kenneth. Fantastic.
Henrik Göthberg:Really good. I love how we talked about our children as well, but that tells us something about so much.
Jeanette Nilsson:Yeah, and.
Henrik Göthberg:Thank you very much, Kenneth. Yeah, thank you.