Aflevering 45: Vooruitblik – AI in het onderwijsjaar ’25-’26

Welcome to the last of the week's A.I.-tests of this year. Your weekly update about A.I. in the understudy. We are Tom and Marcel. Two students who don't only talk about A.I. but also really fit into the class. Today, as announced, we are going to talk about the preview, especially last year. What is going to happen next? First question, which model do you think will really hit next year? Let's go straight into the main topic. Just a quick question. Okay. Well, let's talk about an outsider. I am curious about Grok. Because I have been hearing Elon Musk for a while talking about the new version of Grok. And he says very big things. He does say more, so I don't know if that will really come out. But who knows, is that something? Stop. Okay. Back. Back? This morning, Grok IV came out. Oh, we are already there. Yes, we actually wanted to talk about the future, but the future is here. Yes, we are already there. Yes. I have been here this morning on my way here. And we will take this Donald up until the day before he comes out. After the live stream, I was listening to Mr. Musk in his own way. With a few of those engineers, he has told me what Grok can do. He is a crazy guy, but he is a smart guy. Yes, who sometimes says very stupid and rare things and does. Yes, yes, yes. But Grok IV is much, much stronger and better than all the other models that we have now. And that is remarkable, because the progress... It is the best. It is the number one model now. Yes. Yes. And we don't just have to go to Benchmarks. We will learn the time, but it really seems to be the smartest model. It really seems to be a good big step again. And we actually wanted to do this in the main theme as a future. So we have to keep fantasizing about what will come next. And then we will do it later. Yes, absolutely. And you have to bring it now. I would like to talk about NETN, MCP and Automatization. Okay, clear. I will start with the Automatization. You can do that via NETN. That is a website where you can apply all kinds of automations. NETN is just an N and then an 8 and then an N, right? Yes. And maybe you immediately ask yourself what kind of thing you want to do with Automatization. Well, I have one news case where I have been thinking about it for a while. I just want to build a simple website. You know we can build a good five code, so building a website is not a problem anymore. Where teachers can go. Then they can just do all the tests that have been made by a teacher or student to upload there. Put the answer model down and then they will get back to work. Within ten minutes or something like that. That's all I wanted to do for a while. Automatize, evaluate. Exactly. Well, the time has come to give my private lesson to the listeners of the podcast. I still found a lot of manual work. He said everything you want, you can do it now. But he can do it very well. I really can't do it yet because it's pretty complicated. But I can do it. Yes, probably I can. Then I thought, but I actually don't want AI to make all that automation. I don't want to do that by hand. Exactly. Well, we first went with the normal AI chatbots. They actually couldn't do that well enough. Then I went to YouTube, of course. Because I wasn't out of the field yet. Then we couldn't build something that we could still automate. Then I came across MCPs. MCPs, you probably also hear a lot about them. The people who are in this world are constantly talking about MCPs. They are sort of connections that you make with different things. Where, for example, you can also make an NHS connection within your cloud AI application. That sentence is as cryptic as it is according to me. Essentially, just a chatbot can talk to an app via such an MCP. What you actually want is to say to an AI in normal Dutch style. I want to automate this, just make it. It will come down to that. That just didn't work. So I set up my project for a few months in the fridge. I think it can be done for sure. Let that be clear. But then it costs me a bit more work and energy. Over a few months, this will cost me almost no work and energy. And then we're going to offer this. Maybe right after next year. I will offer a website where you can just let everything go through AI. And you say a few months, but those are just human months. Human months. Ten A.I. years. Let's say full school this year. Okay. That's cool. We wanted to keep the news limited. Because we are actually going to speculate about it. But we do have some best practices. We have to pay attention to that. Such a beautiful end of the year. Talk about some cool things. Shall we start with you? Yes, I think we have had a few weeks about the app I built. To be able to build your own custom chatbot as a teacher. Where you define the knowledge, the didactic and the pedagogic. And then the chatbot will behave as you want. And I can also make a link there. So that your students can use it. I was very proud of that. That's cool. And you came last week with the Meta app. Which you can use as a teacher. To actually build an interactive teaching environment. You were also very proud of that. And also to the right. And right after that broadcast we were so happy with each other. But we have to share this with each other. Why is my chatbot not in your app? Or why is your app not in my chatbot? We played with your app. And then you went further. And I thought, I'll just start again. And I'm going to do that. So it's parallel to playing a game. With who can come up with it. And what we have now. I will take my example. Is that you have an environment where you as a teacher. Can click on which apps you all want. But the chatbot is actually a bit the basis. And it's actually integrated everywhere. The sound is still abstract. But you make a chatbot based on your teacher. And you instruct them to not behave as you want. That was my app. And then you want, for example, a presentation with it. So that the student or teacher. In that interactive teaching environment. Not only can click. But also by a presentation. Can charge over that material. It just automatically makes it. And if you have a question. Then you can just put the chatbot in that presentation. And say, put this out. And if you think. I have something to add to the presentation. You can do that too. If you want to charge that too. And that kind of thing. Can now all with our. Nitwit code, five codes. Modus. So bizarre. You just went in the preparation of this podcast. Let's see how far you are now. Because we were all here. Nice to go to the battle. It's unbelievable what we are already getting out of it. So bizarre. Yes, I think it's very cool. It sounds pretty abstract. I'm going to show it to some colleagues later. And then we can test it for example next year. Because we will come back later. But I think this is really the future. Yes. Yes. Nice. Nice bridge. I think. We are completely surprised. About what we can program. What we can program. I was this week at the High School Utrecht. At the department of ICT. An associate teacher at Dergelijke. She had a five code in the morning. Where I can also show what I can do. But they also had a few who were there. Who could already do very cool things. Five code in the morning is also very cool that you do that, right? Yes, it's really very cool. And there were still a lot of teachers who didn't have that yet. They also experienced that. So we will see a few examples here. But then it was also said. Hey, go try it yourself. And it was facilitated yourself. If you need an account for a month. You can do that. That facilitation was also well arranged. Then there were a lot of teachers. Wow, we are already so far. And if Tom brings this non-ICT teacher. Who doesn't know anything about code. Can build all these things. Then they suddenly felt. Hey, what Tom and Marshall have been saying all the time. Are the knowledge and skills of the future. Are other knowledge and skills than what we are now learning. That's really different. They felt that then. The short film. The short film. And then in the afternoon they talked about. Where will our education go? And I wanted to give such a big compliment. That they were allowed to do this. This is pretty strong for such a training. Especially if you were one of the training software development. What I said there. For five years. My sports students. The students there. They can all build apps. Strongly. They can build very good apps. Because they also have domain knowledge. And building that app. What does the AI do for them? So where are you added value? Well, I thought it was so close. That they were allowed to ask that question. That they were going to save money from it. I thought it was a very big example. For a lot of other courses. They were allowed to do it. And you said five years. But that's to be a little bit more quiet, I think. Yes. Because I think that needs a lot less time. Than five years personally. Yes. But we're going to see that again. Yes. But yes. Really a big compliment. This is what has to happen. I think. In this way. Did you have another best practice shift? Yes. I don't know these people personally. But I read this on the community there. It was an article shift by people from the HVA. Harem training. And the title. We have integrated AI in all courses and years of our Harem training. And tests were necessary. You will be happy about that. When people are working on that. So they will look at all courses. What impact will AI have. How are we going to approach that. And they also had some concrete examples of how they do that. How they have adapted their testing. Super concrete. And what I really thought of. You understand it. You are doing well. The whole training says that he is now AI ready. That's what we want. That's where we have to go. What you mostly see is that a number of pioneers are going to the club. And that it will spread slowly. Especially if there is a sort of oil leak. But these people are already there. And we have included that internally. Compliments. They have also trained a lot of teachers. And with all their efforts. How do we make this education ready for the future. They are so much further than a lot of other courses. Nice best practice today. Yes, great. A positive year. That's nice. There is a lot going on. Yes, for sure. Well, Tom says. For sure. What a positive time. Then we will go to the main theme. But before we do that. Of course we want to ask you to subscribe to our podcast. And to give us our like. Visit our website www.aivodocent.com But I think you are most curious about Iveo's. Famous making of the winner. We start with him. [Music] It was an exciting week for us at Iveo. We got nice episodes. Where we could choose one. The application that we chose. Is made for teachers. Teachers can practice with different formulas. Learn a reaction comparison. And solve the problem. This was interesting. Because a lot of naturalists. And computers at Iveo webinars. That there are still too few A.I. possibilities. For their courses. But this app shows. That you can definitely become enthusiastic. With A.I. if it's about these courses. This maker made an interesting app. With Cloat. Where you can practice with these courses. It is a specific app. Pleasant for teachers who need to practice more. Especially with the explanation. And steps that he gives. Easy to adjust for us at the moment. And that's why we chose this app. As the winner of the Iveo prize question. From the tuesday. So. Mats Walloon. Congratulations. We take contact with you. For the usual book package. Thank you of course. For all participants. On our Iveo prize question. This was the last time. That we shared our contribution. Marcelle and Tom. Thanks for this opportunity. Go on. And we hope. Together with you. And Iveo. A.I. on a good way. Can adapt. Also in the distance without a way. Wow. That's cool. So nice done. Yes. Compliments for Iveo. Leandra. Very very nice to see. That she. Our last sponsor. Or their last sponsor contributions. This way we filled in. Yes. Give it a moment. Happy Mats too. And a very nice example. That we get here. That's a real best practice. Actually. Yes. It fits without a reason. Of course. And we are going to go on. At the same time. Of course. With a. Sponsorless situation. Again. Yes. We have the first 35 episodes. Every week. A different joke. About looking for a sponsor. We're not going to do it now. We're going to do it now. With the reception we have now. We are recognized on the street. That we have a new sponsor in no time. So. Just try to report. Try to get in between. Very nice. Let's go. Let's do it. Really? Can't I continue? Of course. Of course. The main theme. We're going to look at what's going to happen next year. And we had the idea of this. The same structure as last time. So let's look at the technical side first. Then we go to the bottom side. And then a personal reflection. Of what we expect ourselves. And before we go to the technical side. I want to mention a tweet. Logan Kilpatrick from Google. A big guy at Google. He said last week. The next six months of AI. Will be the wildest so far. Make your boss manat. Nice to meet you. Yes. They are really going to betray you. There is that wall. Where we can't go anymore. You can't do that much. That's what it looks like. A logical start is the technical side. And then what's coming up. We had said in the intro. Croc4 is out. It's going to be higher again. What actually plays at the time. It's GPT-5. It's been playing for a year and a half. But they're putting that out every now and then. To create a much more powerful model. That would have to come out next week. My prediction. What I'll tell you later. If that thing is not as good as Croc4. It's going to last a little longer. Because they want to fool each other. They really promised this summer. So it's not going to last long. No. I'm curious. And Gemini 2.5 is out. It's already visible that Gemini 3 is being tested. You can all find it below. And I think those are the three most important models. Someday. That will be high again. And why? And what happens then? What's different about GPT-5 than GPT-4 or Odyssey? I want to mention one thing. Which is also very interesting. It's an open source model. They also announced that it will come in a few weeks. If it's good. I'm also very curious. Oracle will be out. It's even on Tuesday. Tuesday. So those are the models I look forward to the most. And then we'll talk about the coming months. Because we're looking forward to the whole school year. Where we actually look forward to. And we'll just come in the coming months. Then you ask what's really going on for others. At the moment we have all those different models. That's very annoying and very confusing. And one of the things that they promised, which I'm very curious about, is that everything will be integrated. So now you have to. Or actually I don't choose the average user at all. He just starts talking. And then he has the 4.0 model of the GPT-5. But you can find 9 or 10 models. Of which the real pioneers, like us, say, but don't know anymore. And especially O3, to be honest. Sometimes 4.1 and sometimes 4.5. But there are still a lot of models. And that GPT-5 model is smart enough to determine what it needs based on your questions. Super smart. Yes, right? And then everything should be better. Then you actually have the best model for it. And I think that the most important update that is going to come is that the models are becoming more and more smart. That's what we've seen today with GROC. All those benches can be high. That thing is becoming more and more smart. That's what I think it needs to be. That it knows better what it can mean to you. That it has a chance to hallucinate what is smaller. So that we are in the right direction. Because GROC 4 is testing 100% on a number of models. I think that it doesn't make any mistakes. Exactly. That's still important for the chassis. Another thing that is still attached to it is that it has multimodals. So the hearing and seeing of the model. For example, that it is really in the model itself. And that that also means that it will be much more natural. And that that thing can also make pictures and videos. With which you can talk. Those are now all a bit different submodels. I think it makes it much more powerful. Yes. If you talk about all that modality. One thing that I think we will get next year. And what will have a lot of impact. And what people will not continue to have. Is that it will always look at it. So I just put my computer on. And it sees my image screen. And I can always talk to it about everything that is on my image screen. This is going to have such an extreme impact, Marcel. And you once wrote a very famous book about it. 1984. Big Brother is watching you. Yes, you are looking at me with nice eyes, if you say so. If I think about how much we can put this into our education. And we often say in our podcast. Our biggest advantage is that education improves. And people can learn better. And they can prepare for work better. And the potential of it. See what that has to do with how we can learn. That is enormous. And the downside? Can I name it? Certainly. It is also fucking scary, right? Yes, absolutely. So Dennis Osabes from Google. And also Sam Oldman. They have said quite openly. We want to make the universal ultimate personal assistant. So he has to make your whole life with it. He is always on your shoulder. And you are now on the screen. That is one form of everything you do. Yes, there is also a very strange side to it. I actually expect a kind of resistance against it. I expect that too. So that thing can not yet say that we are going to use it. I am very curious. Hey, I want to mention the internet. The internet where we grew up. I do not know which version we are already official. It does not matter that much. But it actually means that you give a search assignment. Because the internet, you can actually find nothing. You are totally dependent on Google in almost any case. And what Google always did was put links. And then you went to click on it. And then you have to scroll down a bit. Because the top ones are just bought links. And A, you are still up there at the moment? Now, but then I will come up there. Because the well-known model was not like that. Say a few weeks ago. And then you went to click on a few links. And then you went to read the website. What has already happened and what I think is going to be much more powerful. Is that you no longer go to links. But that A.I. for you goes to those links. And A.I. does that together. And you only read the summary. And now you see that at Google. Do you know that thing again? Overview. E.A.I. Overview is that. That can also mean that they also want to come up with their own browser for it. They have announced that they will come up with their own browser. Plexity already has that. Plexity. Plexity with coming up with their own browser. Yes. That means that if you take advantage of the agency of the models. They can just search for good information. That actually old-fashioned websites that are aimed at people, say, to operate. They are going away. There is a whole other internet on. And this all sounds like sci-fi. But it is going to happen so quickly. I know that we have also discussed this in the Webinar series. With the last Webinar series. We talked about the new internet, how it is going to look like. And it really sounds like sci-fi. But it is much closer than you think. I really liked hearing how he thought about it. I am so curious how it is going to look like. Yes. I think that if you have a website where you are dependent on people coming up with it. That you really have a problem. You can already see that. So news sites, for example. Nu.nl or something like that. They have much fewer visitors. Because if you ask the chat team what is happening, they go to nu.nl. They get information from there. And you get exactly what you want to have. And your language, say, gets shot in advance. So that is really going to change a lot. I think that is interesting to mention. Certainly. Other things. Are there also things where we don't have such high expectations? I would like to mention one more thing. What I think is also negative. Those models are getting better and better. We are always the enthusiasts. We just want to model as smartly as possible. Because it is always going fast. But we see a sort of weapon law on better models. But also on more expensive models, it seems. Over the top four models that came out, the most expensive model there is 300 dollars a month. And according to me, it went from 0 dollars per month to the first, 3.5, to 20 dollars per month. And now it is already 300 dollars per month. And the free version is limited. And then you get a somewhat less intelligent version. They are still harshly intelligent and free. But I think it is also a bit scary that we will not get a sponsor later. Then we just can't pay those models anymore. That is a problem. Inclusion is really a thing. If we look at education, at this point it is all pay to win. My students, they all have a group that pays the version. Because they get much more. Later you will get even more from the 300 dollars version. And as a teacher, we have to know what the students can do with that paid version. That is a crazy world. So it is a problem. I think it is a serious problem. Really. I think it is the most important thing. What did we miss last year? We said that we had not seen the Vibe Code coming. But I think it is going to be even more powerful. And it was not for nothing so enthusiastic about our interactive teaching environment and so on. What is going to come more there? And I would like to ask you, Tom, Oracle Tom. Is there going to be something new that we do not see now? Or is it just more of the same but better, faster, cheaper and more expensive? Let me put a note first that Vibe Code has a lot of impact on education. Really a lot. And we have not seen that coming last year. And yes Marcel, this is just going to happen again. So we can not really estimate what is going to happen. With something else you mean? With something completely different. But what then? No idea. No, right? No, I mean it's really serious. I think seriously, we can not really estimate how fast everything is going. And what that means again for a lot of different things. And that Vibe Code, I really think that this is going to put the whole lesson on the head. And a year ago there was no second that we were thinking about it. There is going to be something that is really going to change fundamentally. And that's just going to come out again next year. Really true. Okay, well I believe you. Hi. But we're going to make it up. And let's go to the education side. Yes, let's do that. Enough technical things. Yes, yes, yes. Everything is getting better quickly. That is the Teco Message. Because ultimately we only discuss that technical side to the impact of the education side. Exactly. And then we have seen how fast the adoption is going. And what you think of it. But how fast is the adoption of AI in education going now? The same thing as last week. Right. Then you just get the same answer back. I think next year we will all go at a low pace. I just see little or I really see a change. I see progress and I see that people are in motion. But I see that the AI is moving even harder than before. So I think next year just go faster and slow down. Do you know what I mean? Yes, that's a little bit about how you see it. I think, look, you can be pretty fast or slow down. If the goal, say what you want to achieve, is also achievable for a slack. So as the end of the year, everything will be slack. That's fine. Fine, then you've done a good job. What I think or think about is that where you actually have to be, is going to go further away. Then the pace of the slack must be so high that you are also close to it. And not that you are getting further back. If the listener understands me. And I don't know so well. I can also say that if you take the smart approach, actually integrates robustly, which makes new techniques quite easy to integrate, then not everything is in their head again. So if you already have the A.I. ready, I don't think that a new model will immediately take off. Or a new possibility or something. But that you can easily integrate that again. That's what I hope. And I think there will be new problems again. Yes. How do you see the side of the students? Do you see a lot of positive things there next year? Well, I think that it goes on where it comes from. I think that there is not much else going on. And that actually all students go to Chattiparty on an individual basis. That the people who have money, say, immediately have an advantage. Because you say that the difference is getting bigger. That's what I'm talking about. And I think that there are differences between students and students who use A.I. smartly and who use it bigger. And that it actually works de-leveling. And that's not what you want, I think. I think that's what's going to happen. All of a sudden. And you still say people who use smartly and not smartly. I also think that there is a group that doesn't use it. That's also going to be behind it. There is so much advantage to that A.I. and if you can use that smartly, you can't just go there. I don't think there are many people who don't think that it really is a niche. That's actually what it is now. And if we then talk about the interactive teaching environment. Because then we talk about teaching with the students. You said, I would like to come back to that example of Mats, too. Yes, but my thought was just when I heard that. You are a teacher and you are at Mats in the class. Mats gives of course, as I understand it. You get that great app from Mats. You learn with it and then the Zoomer goes. And then you walk to the teacher. I'll take a different technical class, say. I follow you. And then the teacher says. Open your book on paragraph 6 of chapter 3. Do you still have that? Don't you say to the teacher, why? I would like to have interactive learning apps. So I think it's also very complicated for schools. If you have pioneers who are going to integrate these kinds of things. Then you also get a question, say, at the students. And I don't see the way they want to go back. Yes, and you are now in the front of the teaching. But we just do this in the daily practice with us. What we are doing with all these cool things. Other people even less. This is what we also participate in. I think so, yes. So I can still see that. It can also be a drive to put it all in a little faster. So it can also be a good trickle. But it's also a strange situation that you can do something at the same time. And not at the other one. Or get it and not at the other one. Yes, and let's continue here for a while. You also have the best practice. Like Dwayne at Anatomy. What was it? Yes. Then you have an AI teacher you can talk to. But as soon as we offer the interactive teaching environment. Then you have the AI teacher next to you. You just told us about a presentation with an escape room. With flashcards. With everything on the table. Yes, that's all in a while. And with the other teachers, they will say. Take your book with you. Yes. Then the student will be crazy. It's not. Or am I crazy? Well, I think that's the way it is. Yes. I think that's weird. From the student's perspective. Yes, I think so too. And that you have a good story to tell as a teacher. And you can do that. But then you have to make it clear. Why are you doing it at the same time? If you are conscious, it's not a problem. Exactly. Absolutely. But now it's still as if a teacher is more trained. And they can build cool things. Then we have the students much more. If they have that teacher. I think so too. And another aspect. When we look at students, MBO, HBO, uni. I get more and more in my example on Twitter. Then I get to see AI replacement jobs. Yes. Really more and more. That's the trend now. You see that now. And that's also happening with Dutch companies. And then not one on one replacement. But you just put people out or not connected. Because the rest can deal with AI much more work. Do you think that the future will have a lot of impact? That we will really see that as a trend? Or is that too early? As a trend, certainly. I don't think there will be a mass strike or something. I just think that the impact will have on the job market. That we will certainly see that in the future. I also notice it on all kinds of things. I was actually dealing with accounting rules. And now I think, no, I'm going to have my own app-vibes code. That will all be regulated and automated. Then I immediately think, yes, that's pretty intense. I gave a workshop to experts. I think if you keep doing what you keep doing, then you won't be there anymore. Just very close to what I'm all involved in. And I also see the same news feature that you see. That there are a lot of people in different areas of the field. And certainly in America, it will be much faster than what we see here in Europe. Yes, it's a trend, not too large-scale yet, but yes, I see it. Okay, hey, docenten, next year. Everyone is taking a step forward, I think. I think there is a very big need for education. I hear, when I walk into the steps of education, we come across a lot of different education. Above all, we are now talking about AI. And I think that's very good, because then people understand a little bit how much impact this is going to have. So the need for education is enormous. A lot of people have it. I know that I have to do something with it, but I don't know how to start. I don't know how to help. Do you really see that there are steps being taken into account? Yes, we have often said that. I notice exactly the same thing. And I think that last year, also because you are going to see a lot more examples in the practice as a docent. You think, here I have to do something, and here I want to do something. We feel that trend in our network. And then I see, say, the low-ranging free docent, a workshop, how do I work in chatbots, a vibe code in the morning, but we really want to have something integrated as a goal. And then we hear a lot more steps in a little bit of a school. We see that too. Can I plug the webinar series from you? Scameless. Scameless. Without any scameless, but we make impact. So I think that's cool. But this is a half-year project. And there are not only the webinars, but also the Q&As. And we give you apps that you can use. And yes, it is scameless at the same time. But seriously, this is one of the best courses you can get right now, I think. I think so. I think we have put together a very powerful series with Barendt. Where you really, well, where you say to me, what did Barendt say? You learn as a docent nothing on a study day. You learn between study days. And I think that's the strength that we make sure that those 10 webinars with everything in between, that you go to the battle. And that's really something else than a workshop you follow. Enough about us. And one last thing I want to say about the subject. I don't want to say any more scriptions. We really have to think about that. As in "Gronen tootsen". So I see that a lot of people are working on that. So it's really going to be on the agenda next year. It's really going to be on the agenda now. And I'm very curious to look at it. I said at the beginning that I could just build something where the AI will look at it. And already know what can be very reliable. That's going to have an impact. How happy can we make docent here? Well, I think that's the most heard-of workshop. Yes, absolutely. Well, administration as well as numbers and so on. But that's as old as education. You have to write in your letters about the agenda. And now you have to go to one of the online systems. But the assessment costs a lot of time. And if you can win with time and the quality can also win. Then that win-win. I do see things happening on the left and right. We see some pilots coming, so that's going to happen. Actually, it's important that we're going to see a lot more in practice. And of course we're going to be happy about that. So we're not only going to talk about that, but the practice is going to change. It's going to do things. Yes, come on up. I still want to say something. What I see happening already, I think you're on your own. And what I think is going to happen even more seriously, is also at the end of that call last week from the academic side. Yes, that in my opinion is really a ridiculous assignment to get my AI out of education. Yes. HBO and Uni. And that's mostly the universities that are anyway a bit lower in my experience. The integration is still in the expected position. Or even in a negative position, you shouldn't want to do that. And of course those students don't use that Dominic at all. And I think that the universities are still going to go back. I think so. And that actually, what do you have to say right away? Because that was always the place where you really enjoyed all the major education. Yes, that's not what I'm talking about anymore. That doesn't matter anymore. They can't keep up with this position and this doesn't matter anymore. Yes, but I still have to see it. I still have to see it, but it's not possible that they are still going to ban this for another school year. It's not possible. Yes. You almost say, in my opinion, that last year the gap between the universities and other education institutions is even bigger on this topic. Prove me wrong and I will let you talk to me. I do believe in that. That they will make the environment. Well, that's enough about the education side. I have already said a lot today, I think. I think that we have mentioned the most important things. And it's all about speculating sweat in the room. That's right. Yes, yes, yes. But yes, from two oracles, so to speak. That's right. Hey, shall we let the last topic, the personal reflection, go? So what, how do you look at that for the next school year? Yes, I mainly, it was really a tough year. Because of all the races, but we just played for quite some hours, say, right? Yes, until then. And I notice that where I got energy from this year and why it was possible to work as an idiot for many hours a week, is mainly to make that impact. And that I just see in the practice that students or students learn better or faster and that teachers get grip on it. And that that is the positive side of the story. And I really look at, I call it a reverse pyramid game, that if I can make an impact by giving a number of teachers a handle, that those teachers are going to influence their colleagues again, and that the game is going to go like this, say, it's a positive pyramid game, I think that's really cool. Why do you call it a reverse pyramid game? Is it a sort of oily area or something? Yes, a strange thought, I don't know if I'm making that up. Okay. And I am, and that is, yes, there is also a question. I am of course also just a teacher here at my own education institution. I think I'm a pioneer in the sense that I think I walk forward in what is possible and have a lot of ideas. For example, my Meta app that I was later named. I just wanted to test it with us. And my question is, and I'm looking a little bit at that, how far is it possible to test my interactive teaching environment later? And that I don't pay the API costs, but that we, for example, use the pot of your teachers, what we just have, say, that we can use that for it. Because there are still some gaps between people who might find that exciting or something. What do you think? How sustainable is that? That's a good question. And if you really want to do it at our level or at our education institution, it's almost impossible for me to feel that fascination within our academy is, I think, as much as possible. So if you can rule within that, then there is something. Okay. Well, that doesn't sound right. I want to end positively. I think there is more possible than we think. There is more happening within the education organization. It's just a big ship. And I'm going to look for my ways. I'm going to see where the haze is going. I'm going to look for a way to experiment. I'm going to see where the haze is going. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment. I'm going to look for a way to experiment.