He is considered a "rock star of mathematics" and inspires and supports countless students with his YouTube videos for mathematics: However, Daniel Jung is much more than a YouTube star. As an educational entrepreneur, he is committed to the transformation of the educational system, educational equity and a change in learning.
In this context, Daniel Jung also deals with the topics of artificial intelligence (AI) and machine learning. In an interview with Gwendolyn Paul from the zdi regional office, he asks himself the following questions: what needs to change in the education system so that we can use the advantages of AI? What skills do we have to teach young people in order to be able to use innovative technology responsibly? And what role can extracurricular educational opportunities play in this?
We conducted the interview on April 27, 2023. You can watch a summary of the full conversation here:
AI in education
Gwendolyn Paul: The topic that interests us most in this conversation is the topic of AI in education and above all extracurricular education. What is your opinion: At what age should one start with extracurricular offers on the subject of AI? Should we make it possible to try things out and get to know them as early as possible? Or are you more of a stickler for the classics Career and study orientation from the seventh grade?
Daniel Young: There isn't one answer. But we should certainly get our "offspring" used to the world as it works now at an early stage. And if an important part of this world are robots that act in a network through artificial intelligence and also probably intervene in real life, then you should have experienced something with robotics by the seventh or eighth grade. If that isn't guaranteed at school, then we have to do it somewhere else. Because the world around us has been reorganized. So we must act now.
Gwendolyn Paul: What do you think would have to happen for that? In a video you made the demand that politicians should let more money flow so that schools or universities can implement or test AI more.
Daniel Young: In my podcast I have conversations with different guests: professors, teachers, entrepreneurs, people from politics. I get the message from them that politicians really have a hard time implementing things. Even if you have five million to spend, the questions come: When will they flow? Where are they going? In order to be able to act as an economic nation in this transformation that we are currently going through, we need implementers and designers. Designers who then meet courageous headmasters and teachers who say: There are ways to get to grips with robotics in a playful way. Then let's take advantage of these opportunities!
What are the most important future skills?
Gwendolyn Paul: If we look at a course or subject where AI-enabled learning tools could be used, what impact could that have? On the one hand on the skills that you need, but on the other hand also on things like personnel. So it could be that I need a lower supervision key because the AI takes care and less personal supervision is needed? How would you assess the impact?
Daniel Young: I think there are still many misunderstandings. Many see AI as something with which they can now learn completely by themselves and everything is optimized. I see AI – especially at school – more as something that can create free time. For example, AI could be used cleverly to simplify administrative processes for teachers. They would then have more time to do something with the students: for example, to find out what artificial intelligence and machine learning are. Otherwise ChatGPT is simply tested without knowing what is actually happening. And then there is the syllabus that we have to stick to and in two weeks we have to do another math test, then one in German and one in biology. And the knowledge about ChatGPT has already fizzled out again.
What does the abbreviation GPT actually mean?
The abbreviation stands for the English term "generative pretrained transformer". The term describes a specific type of language model that is trained in advance using existing texts.
Learning with the help of videos is optimized and leads to faster learning success. If you use this skillfully in combination in all subjects, you could create a day of free space on which the students could then attend a robotics course, for example. To which someone from business could be invited. Or where you can talk about AI. I think the potential is still not properly understood. Because we cannot change the curriculum overnight. We cannot bake new teachers and cannot implement new school buildings. But freedom can be created with the support of AI.
Gwendolyn Paul: If we now take the step out of school: In extracurricular learning opportunities you are much freer, both in terms of the learning plan and in the organization of your time. What would you say are – outside of the school structures – the skills that need to be taught?
Daniel Young: After many years of Google, you said: "What do I still need to know? I can google anything." But that's the absolute myth, because it's just not like that. I need a foundation, I need to get the neurons firing, I need to be able to do research. I need to know: What is the source reference? This is a huge topic in AI and perhaps an important first skill. Topic Deep Fake: Do I know that this was really a video of the President of the USA or was the video AI-generated? Am I able to question things, research them and weigh them up with experts and communicate: Was it really real?
Perhaps one of the greatest skills is to question things and then to research what should have been done in Googlefication. How can I legitimize that information is really correct and how do I get it.
Gwendolyn Paul: So a certain level of media competence is required.
Daniel Young: It used to be called media competence and in this sense it is still media competence today. But today it's brutally different to see something in photos and videos and no longer know if it's real. And so you come to the things that will be important for the coming decades. What does lifelong learning mean? I should always be able to teach myself new things, even alone and in a wide variety of formats such as analogue, digital and supported by AI. I should always question things, should ask questions at all, be able to take criticism, discuss and communicate. And also digitally.
Then we come to issues such as communication, which is essential as a skill. I come from mathematics and I celebrate mathematics. But when I have understood a derivation in principle, I don't have to do ten more derivations, because it is already understood. And when I understand things, then I no longer have to learn them because I have understood them. And that's when you start to ask yourself: what should I be able to do as a responsible citizen in order to find my way in the world?
A complex problem for many is to consciously teach yourself something new. Many do not even see the need to do anything outside of school. But now we're realizing that advances in artificial intelligence -- or rather, machine learning -- are having incredible implications. So we have no choice but to use extracurricular projects and initiatives. I am responsible for drawing attention to this on the internet and for creating the necessary freedom with all my projects. And then we might be able to draw attention to these skills and convey them.
Which industries is AI changing the most?
Gwendolyn Paul: A certain openness is then also one of the future skills, to put it in a keyword. In which sectors do you see the biggest changes coming our way?
Daniel Young: In all! But of course very fundamental in education and medicine. These will be two areas that will experience an unparalleled revolution in the next five years. The world doesn't really care whether we in Germany set up our school system for the future. And meanwhile, for example, large modern companies build their own schools and universities. I think it would be a shame if we weren't on the road creatively from Germany. And if there are already many projects and initiatives, then you should bundle them together. One should say: Hey, you Teslas, Metas, Googles, Open AIs and Microsofts of this world, let's join forces here in Germany! We can do something too! And that's why the question now arises: How can extracurricular initiatives with PS reach everyone so that they can experience and test it?
What is the AI Learning Assistant AIEDN?
Gwendolyn Paul: I would like to at this point AIEDN enter, the AI-based learning assistant that you are currently developing as part of a research project. Can you tell us something about what you do there and when you can see something of it?
Daniel Young: First of all, this is again an extracurricular project. Although it would be so nice to be able to implement it directly in school with all students.
I've found that you can break many pages of text and days of instruction into a sequence of "five minute nugget learning units." And that you not only learn with these units, but also understand them. Then I asked myself: Do you really have to watch the whole five minutes? Because the analysis has shown that people only watch for about two minutes at a time. If you now imagine that you could reduce the time again. One could consume units of knowledge at the appropriate point in the video in such a way that one understands it.
In a nutshell, this is what we want to make possible with our AI-based learning assistant. The first field study is already underway with students who make an entry and are immediately directed to the right place in the video. The study is also accompanied by neuroscientific research and we therefore know that we not only learn, we also understand.
Gwendolyn Paul: That means the students ask a mathematical question and get an answer?
Daniel Young: Exactly. Classic example: what is a fractional rational function? If you had to answer that now...
Gwendolyn Paul: ...then I would be happy about an AI!
Daniel Young: Then you would probably post the question on YouTube and watch a few five minute videos and eventually you would have your answer. You enter the question in our prototype and get the appropriate 30 seconds, maybe another 30 to 40 seconds, and then you can say: I understood that and I can repeat that. If you imagine this principle not only for math, but also for physics, chemistry - the content is there - then we create an incredible amount of time. Because you no longer learn bulimia, we accompany the learning in such a way that you really understand it.
How the learning assistant can optimize learning and understanding
Gwendolyn Paul: Isn't that an incredible abbreviation of knowledge? You don't take the path with you at all.
Daniel Young: If I'm studying a complex mathematical topic and I just don't know what a power is, then I'm missing a crucial link to get to the aha moment. And then suddenly there's a 30-second sequence explaining the potency. Then you can tick a box, you have understood it and can use it.
It is my vision to create free space for learners and teachers at the end of the day, who can then tackle other things. In the next few years, a lot will happen at ever shorter intervals, and at the moment learning and understanding is not optimized.
Gwendolyn Paul: I have two more questions about AIEDN. When can you go out with it and who can use the learning tool? Will it be a freely accessible tool or is it only free for certain audiences?
Daniel Young: As part of the field studies, AIEDN is currently being used in cooperation with a university and internally at schools. We plan to make the tool available to everyone in July/August. I am fighting for it right now and will not be deterred from making the tool available as open source for pupils, students, parents and teachers. That's the democratic approach, you shouldn't pay money for it anymore. The tool is a kind of marriage of my platforms, which are currently free anyway. This installation of an AI learning assistant is not done overnight, but August is realistic.
A little look into the future
Gwendolyn Paul: I have to say, this is an exciting project! And also a building block on the way to changed learning or new possibilities of changed learning.
Daniel Young: Yes! Maybe we will build a digital universe from Germany where you really only find good content. And instead of getting lost on TikTok and suddenly only looking at vegetables and junk for an hour, there would be an alternative. Then you would have learned a lot in math, physics and computer science instead. Maybe I'm a dreamer there.
Gwendolyn Paul: I think that's really exciting and we'll definitely take a look at what happens with AIEDN in the summer! Is there anything else you really want to get rid of?
Daniel Young: I've actually already said it: We need courageous designers now. Let's go!
Gwendolyn Paul: Thank you very much for the interview, I am very happy that it worked!