05.08.2026 Sabrina Ludwig 150

Episode 51: AI in Schools: Are We Still Learning on Our Own?

What can AI really do to help with learning, where do problems arise, and how can parents and teachers support children without having studied computer science themselves? This podcast episode features podcaster and education influencer Bob Blume, known to many as the “Netzlehrer.”

Here's what this podcast episode is about:

Bob Blume is a high school teacher, education influencer, and SPIEGEL bestselling author of "Warum noch lernen?" (Why Learn Anymore?). He advocates for a school system in which learning and equal opportunities are more important than pure subject matter and grades. With a large social media following, he raises awareness of educational issues and brings the classroom perspective into the public debate.

Excerpt from the podcast interview "AI in Schools":

Oliver: If AI were a student, where would it sit? Right up front like a nerd, way in the back like a slacker, or in the middle with dangerous half-knowledge?

Bob Blume: Right in the middle, with dangerous half-knowledge. That’s also the real point we’ll come back to later. It’s a good question, but the most important answer is: AI isn’t human. So when we talk about AI, we always have to ask: Where’s the human element? I think we should keep that in mind.

And it has been proven numerous times and in various ways that AI hallucinates—that is, it makes things up, such as researchers who don't exist. On the other hand, it's half-knowledge, which means that, to some extent, it works very well.

I don't want to downplay the potential of AI. Even if you've never used it before, you can perhaps think of it as a kind of powerful amplifier: AI amplifies what someone was already good at.

So, anyone who can evaluate the AI’s answers can certainly continue to grow with its help. But you can also learn significantly less if you rely too heavily on the AI and end up deceiving yourself. That is precisely the problem with this not-quite-nerdy, half-baked AI student sitting in the middle of the class.

A Concrete Example: Biotest and AI as Learning Partners

Oliver: Let's take a biology test in 7th grade on the topic of digestion: How could we use AI to help the student learn?

Bob Blume: If we take this biology test, there are probably a lot of notes and information that you have to look up in the book. Then you tell the child, “Sit down and study.” But that’s not really a strategic approach. 

For example, I could feed all the content to the AI and say, “Use this to create a study plan for one, three, or five days.” I could also ask it to create a multiple-choice quiz based on that content. That would be the simplest use case, but suddenly I’d have a study aid.

There are programs such as NotebookLM from Google. In everything I’m about to say, I’m excluding applications that do not comply with ethical and data protection standards. That’s something else to keep in mind.

In NotebookLM , I could save my notes, information, and sources and say, “Create a mind map from this” to visualize the content. I could also say, “Create a podcast for me.” Then, suddenly, two people would be talking in German about the topics from my biology test. One voice might say, “That’s quite a lot here.” A female voice would reply, “Yes, that’s true. But let’s start with the bones.” In theory, the child could then review the entire material during a 15-minute walk. There’s even the option to have a video created.

A third level would be a dialogue with AI. This requires powerful AI. For example, you could give it the command: “If I don’t answer a question completely, please tell me that my answer is too vague and ask again.”

Then, as a learner, you’re sitting there and suddenly have an ideal learning partner— ideal not in terms of humanity, but in the sense that the AI always meets you right where you are. And none of this is just a pipe dream.

Oliver: How can teachers use AI effectively to provide students with more individualized support and better guide their learning processes?

Bob Blume: These are known as adaptive tutoring systems. By automating certain learning processes, AI can do something that teachers are often unable to do: It can assess, at any given moment, where a student is in the learning process and what he or she needs right now.

By the way, I did that myself as a teacher, specifically in my German class. I used a program that provided students with feedback on their written work. I had created the categories beforehand, and they were clear to the students.

This is so interesting because, as a teacher, it allows me to take on a different, very meaningful role. For example, the student receives feedback like this: “Your argument is very well-reasoned, but you should work on approaching the topic in a more nuanced way.” That’s helpful feedback. Otherwise, I wouldn’t have known how long I would have had to read to realize that Marco is having trouble differentiating between arguments.

Teachers usually give this kind of feedback only once—during a class test. By then, however, it’s too late. The key point is: Maybe Marco doesn’t even know exactly what “differentiation” means. So, as his teacher, I go up to him and say: “That’s interesting. Let me quickly explain what that means. Now try it again and revise your text. I’ll come back to you afterward.”

This suddenly gives me the opportunity to identify much more precisely where each student's difficulties lie and to take targeted corrective action.

Homework Is Dead – AI and Fairness

Oliver: In your opinion, how is AI changing homework? What would be a good homework assignment that uses AI?

Bob Blume: Homework is dead. I said that on January 23, 2023. Three years later, the president of the German Teachers’ Association said we might have a problem with homework. His solution: Students must write their homework by hand, because then they can’t cheat using AI.

Homework has always been a problem because it shifts the learning process to a setting where people have different starting points. Either you have a father or mother with an academic background, or you have a tutor—or you don’t. And now you could say, either you have the best and newest payment model from Claude, Chetchipiti, or Gemini—or you don’t. That’s a problem. 

The central thesis of my book *Why Bother Learning?* is that learning should be at the heart of school. Because when things get difficult, we can’t just say, “Now do this as homework.” As teachers, we need to be there to support them.

Oliver: How can parents help their children learn to use AI effectively? If I had 15 minutes with my child to introduce them to AI—what should I do?

Bob Blume: In my parenting book, *How Can My Child Succeed in School?*, there’s an entire chapter on AI and another on media literacy. Media literacy isn’t just about technical questions like “How does this work?”—it’s also about adopting a certain mindset.

Trust is important here. In my opinion, the most important skill for parents is showing interest—without judging, at first. Many children may already be quite advanced, or may think they are. Many adults, on the other hand, feel they should already be further along. If they aren’t, they feel embarrassed and would rather not pursue it.

It’s aboutshowing an interest in children’s media worlds: “Why are you doing that? Please explain it to me. I don’t understand.” Of course, you can ask critical questions if you don’t approve of something. That doesn’t mean you can’t set any limits. The key is that we, as adults, don’t immediately judge everything based on our own upbringings.

When it comes to AI, it’s also about curiosity and transparency. You can experiment together: “Let’s see what the AI comes up with. What happens if we ask it, ‘Can you explain this task to me better?’” If you have time for that, you should reflect on it together—even if that’s a big word. 

Basically, it just means asking yourself: Does this help me? Does it make me better? Or does it just make me feel like I can do something, even though that’s not actually the case? It’s a matter of interest and trust.

Oliver: Is there anything that children are intuitively better at than adults when it comes to interacting with AI?

Bob Blume: Yes, I think it’s about being naive. There are adults who are hesitant to try AI because they think, “I’ll have to spend a lot of time learning how to use it first.” 

Children, on the other hand, often approach the matter with a more open mind. They think, “I’ll just ask a question and see what happens.” Depending on the answer, they ask another question.

Sometimes I compare their approach to AI to Dobby, the house-elf from Harry Potter. Children approach things intuitively and don’t feel like they can do anything wrong. That’s a huge advantage, especially when it comes to new technologies.