Introduction
Welcome to another episode of TJ’s Technology Tuesday. And you may notice things look a bit different today. Why? Because I am in a hotel room. I am delivering an in-person presentation today, arrived last night, and I am giving a seminar from my hotel room during the break right now. Of course, I did not want to skip the TJ Technology Tuesday impulse, because every Tuesday at 12:30 there is a live impulse — or, as in this case, a short recording made just beforehand, because we are continuing shortly afterwards.
Does AI Make You Stupid – Really?!
The topic for today’s TJ Technology Tuesday is: Does AI make us dumb? Last week there was a lot of hype in many posts commenting on a study that had people write essays — 20 minutes per essay. One group wrote without AI assistance, the other with AI. Brain activity was then measured, and the conclusion was that you need significantly less brain power when writing with AI support. That is not particularly surprising. I find it quite logical, actually. But why is that? It is a tool — and just as using a calculator will result in fewer synapses firing compared to doing the calculation manually, this study design is, in my view, not entirely sound. Setting aside the fact that the sample size was only 18 people — representative studies typically have much larger samples — the more fundamental issue is that the wrong thing is being measured. The question is not: if I use a tool, do I need less energy? Obviously yes. The real question is: if I rely solely on a tool, do I unlearn how to do things myself? A classic example: if I use only a calculator, do I lose the ability to do basic mental arithmetic? To some extent, yes. Back when calculators were not allowed in school, students had to do much more mental calculation. I can still remember my wife’s grandfather — he ran his own painting, decorating, and wallpapering business after World War II. With a decorator’s pencil, he would quickly calculate on the walls exactly how much wallpaper he would need.
He could do that in his head, he could do it by hand. Those are skills that are, to some degree, being lost. But the idea is this: when we introduced calculators — I believe it was from about the 9th grade for us — the freed-up mental capacity was meant to be used for more demanding mathematical problems. Our daughter took this to the extreme: she studied mathematics for her bachelor’s and then her master’s. If I ask her, “Nina, can you do mental arithmetic?” she says, “No, I can’t. In our maths department, we did not calculate — we thought about abstract concepts.” And that helps her today as a programmer and AI developer to think in highly abstract terms. You also have to consider — there was just recently a major power outage in Spain and Portugal. I saw a wonderful segment on television about it: someone handed another person a map because the GPS systems were down. And then they looked at it: “Where am I on this map?” It was quite a different experience from what the previous world had looked like. Sometimes you do unlearn certain things, but you gain others in return.
Now there is an important point to make here. You only become dumber when you use a tool mindlessly, without thought or understanding. If I do not know the order of operations when using a calculator, then even the most powerful calculator is useless to me. If I enter the wrong street into a navigation system — say, in Berlin where the same street name exists four or five times — then it will take me exactly where I told it to go, but that might be the wrong address. So the best results come when you combine HI — Human Intelligence — and AI. That is exactly the idea: ideally, the two multiply each other. And if you have 0 times 100, you still get 0. The more human intelligence you bring to the equation — that is one side of the multiplier — the more AI can amplify it. This is an equation where both sides are essential. The same applies to learning. There is always a big debate about this. Just yesterday I read a very interesting post on the topic — it described how learning with AI can now be supported through self-made videos.
What was interesting was that most people did not even read the post to the end — they assumed it was about teachers making videos and simply telling stories well. But the post was not about teachers producing history videos at all. It was about students — much like they used to prepare presentations or make collages — now perhaps creating a video with AI help in which they tell a story. And that playful instinct is something you can harness, which makes it a worthwhile approach. When it comes to learning, then, when is AI not a good idea? AI is not a good idea when it means: “Write me a homework piece on this and that topic” — and you simply hand that task to ChatGPT. That is not a good idea. A good idea, however, might be to say: “Here is my first draft.” Upload it and then say: “Please revise this and make it even better. How can I improve this text?” Then you get suggestions for improvement. And that has the advantage that, firstly, the grade will be better because you have run one more revision loop before the teacher corrects it.
And on the other hand, you also get an extra sparring partner beforehand. That is how it should work. Not on autopilot, simply entering a question and getting an answer back. On the topic of questions and answers: another good idea — not only but also when it comes to learning — is not just to ask the AI questions, but to say: “Dear AI, please ask me questions instead.” You could use AI to prepare for exams by uploading a set of lecture notes and saying: “Please ask me exam questions. Ask me one question, evaluate my answer, and then ask me the next one.” The good thing is that if your answer is poor, the AI will generally not say: “You are completely clueless.” Instead, it will typically say something like: “That was partially correct — please take another look at these specific points.” AI wants us to use it more, and for that reason it tends to be encouraging. It provides positive feedback. That is a really good approach when it comes to learning.
Learn Faster with AI
Another great way to use AI for learning is with Google’s NotebookLM — you can upload all your learning materials there: lecture notes, your own handwritten notes. You can also add YouTube links, websites, and texts, and then have it generate an approximately ten-minute audio podcast from the material. What is great is that you can have the output in different languages. For example, if you have a school-age child who has a physics test next week and an English test the week after, you could have a podcast on the theory of relativity prepared in an age-appropriate way — and have the whole thing delivered in English. Two birds with one stone. So it is similar to the calculator example, or the Google example — many people said Google would be the end of education. Well, if you do not know what to search for, it may not help much. The same logic applies here. Rather than using Google or ChatGPT for research, I would always recommend Perplexity. Perplexity is built specifically for research: it allows you to look things up with source attribution and attach additional information as you go.
Let me share two or three more ideas on how to use AI as a learning booster. For instance, if you are watching films on Netflix or videos on YouTube, you can enable subtitles. There are add-ons that let you display subtitles not just in one language but in two simultaneously. Our daughter, for example, had a Japanese test just yesterday and had prepared partly by watching manga videos on YouTube with subtitles — in Japanese and German (or English). She could read along in Japanese and also see the translation. So the next time your child asks, “Can I watch another episode of my favourite series?” you can say: “Absolutely — but watch it in English, not German, and turn on bilingual subtitles.” On Netflix, you can very often switch to the original language. Using technology intelligently — that is a good idea. AI is also great for language learning — and I am a big fan of smartphones for that. I am not in favour of demonising smartphones.
I am in favour of using smartphones intelligently. And often, the parents need to break their own habits first. If the parents are glued to their screens at the dinner table, it is no surprise that the kids copy them. What is interesting here is the voice mode in ChatGPT: you can activate it and say, “You are my English tutor, your name is Tom.” And then: “Dear Tom, ask me questions.” You could say, “I have an exam on the theory of relativity next week — please ask me exam questions about it, suitable for Year 9, a Gymnasium in Bavaria. Do it all in English. I will answer in English and at the end, give me two pieces of feedback: first, how well did I answer the physics question, and second, how good was my English? Give me feedback on both.” AI is absolutely brilliant as a tutor for this. And where else can you find someone who gives you immediate, constructive, and positive feedback on two things at once — the subject content and the language quality? At the end you can say: “We have been talking for about 20 minutes now on this topic.
Using AI as a Sparring Partner
Please now give me a summary of the tips you gave me: where should I look again in terms of the subject matter, and where should I focus when it comes to language?” That is using AI as a true sparring partner. And that is how it really works. Or for language learning — Duolingo, for example, which I love. AI is already integrated there, with pronunciation training, fill-in exercises, and playful learning elements. That is why, for me, AI is not the death of learning. It only becomes that when someone says: instead of learning, I just go in and use the tool. When you use Perplexity, for example, you get a question answered and it immediately offers further questions to explore. Those Perplexity results can be fed straight into NotebookLM. For NotebookLM, by the way, I often use the smartphone app. If I find an interesting study I want to read, here is what I do: I upload the 52-page scientific PDF into NotebookLM, and it turns it into a ten-minute podcast in interview format — which I then listen to while jogging, in English.
Since I regularly give talks in English and run English-language management workshops, I also train my English at the same time. And the great thing is that even when I then read the paper itself, having already heard the interview-style summary, I already know what it is about. You may know that I am a speed reader. One of the ideas behind speed reading is not to read a document just once and retain only 10%, but to read it five, six, or seven times. If you have already listened to it once beforehand, you may only need three or four passes through the material and can retain significantly more. And especially with dense material, this can really help you get into a subject. Sometimes you do have to get your head around things like the AI Act, the General Data Protection Regulation (GDPR), NIST, DORA, or any of the other lovely regulatory frameworks that exist. You can upload all of that to NotebookLM, listen to a podcast on it during your commute, and I upload books chapter by chapter — one interview per chapter. With ten chapters, that gives me ten times ten minutes of interview content, and then I only need to speed read the book and have it immediately present in my mind.
Conclusion
So you can see: does AI make you dumb? No, AI does not make you dumb. If you have school-age children, I have developed a brand-new AI course specifically for schoolchildren — created together with Jens Vogt from the LERNcafé. I will put the link in the comments below this video. Behind that link you will find a recording of over an hour from a webinar on the topic of learning with AI, and on the page you will also find a link where you can get an AI course for schoolchildren at a special price — originally 497 euros, now available through the LERNcafé for 197 euros. There is also an AI tools list available — whether you go it alone or with support. What matters is not whether you buy the course from me — what matters is that you engage with the topic of AI for yourself and with your children. Because it is not AI that will replace people — it is people who use AI intelligently who will replace those who do not, in a professional context. That makes it more important than ever, today and in the future, to be AI-ready. And I am genuinely happy to help you get there.
Your Personal IT Coach for executives — and as a father myself, an enthusiastic learning coach for children too — Thorsten Jekel.
Key Takeaways
- A study with only 18 participants examined brain activity when writing essays with and without AI — but reduced brain activity with AI is not proof of becoming dumber.
- AI works like a tool: just as a calculator reduces the need for mental arithmetic, it frees up capacity for more demanding tasks.
- What is decisive is the interplay between Human Intelligence (HI) and Artificial Intelligence (AI) — the more HI you bring, the more effective the AI becomes.
- AI as a learning booster makes sense: for example, have AI help revise your own texts rather than having it write your homework from scratch.
- Google’s NotebookLM lets you upload learning materials and generate an audio podcast from them in various languages.
- ChatGPT’s voice mode can be used as a personal tutor that poses exam questions and gives simultaneous feedback on subject content and language quality.
- Perplexity is better suited than Google or ChatGPT for research-based learning, as it provides sourced results and suggests follow-up questions.
- Duolingo uses AI elements such as pronunciation training, fill-in exercises, and gamified learning for language acquisition.
- It is not AI that makes people dumb — it is the unreflective use of AI without one’s own thinking and understanding.
- People who use AI intelligently will have a decisive advantage in professional life over those who ignore it.
Frequently Asked Questions
Does AI actually make people dumber?
AI does not automatically make people dumber. What matters is how the tool is used: those who use AI without thinking for themselves lose opportunities for mental development, while deliberate use frees up capacity for more demanding tasks.
What did the study on brain activity during AI use show?
The study had 18 people write essays — one group with AI support, one without — and measured their brain activity. The result showed less brain activity when using AI, which Thorsten Jekel considers unsurprising and methodologically questionable, given the very small sample size.
What parallel does Thorsten Jekel draw between AI and the calculator?
Thorsten Jekel compares AI to the calculator: it reduces the need for mental arithmetic, but frees up thinking capacity to solve more complex mathematical problems. The same applies to AI — it takes over routine work so people can focus on more demanding challenges.
How should AI be used for learning and homework?
AI should serve as a correction and improvement tool when learning, not as a replacement for one’s own work. Instead of having ChatGPT write homework from scratch, Thorsten Jekel recommends uploading your own draft and asking AI for suggestions on how to improve it.
What is NotebookLM and how can it help with learning?
NotebookLM is a Google tool into which you can upload learning materials such as scripts, PDFs, YouTube links, and websites. It then generates an approximately ten-minute audio podcast in interview format, available in various languages — ideal for learning while jogging or commuting by car.
How can ChatGPT be used as a personal tutor for exam preparation?
Using ChatGPT’s voice mode, you can simulate a personal tutor: give ChatGPT a role (e.g. “You are my English tutor Tom”), have it ask you exam questions on a subject in English, and receive feedback after each answer on both your subject knowledge and your language quality.
Why does Thorsten Jekel recommend Perplexity over Google or ChatGPT?
According to Thorsten Jekel, Perplexity is specifically designed for research: it delivers sourced results and immediately suggests follow-up questions, making it more suitable for learning and information research than general-purpose AI tools.
How can AI be used for language learning?
AI can help with language learning in many ways: Netflix and YouTube series can be watched in the original language with bilingual subtitles; ChatGPT’s voice mode enables conversations with a virtual language tutor; and Duolingo uses AI elements for gamified language learning with pronunciation feedback.
What does the formula HI × AI mean for the use of AI tools?
The formula HI × AI stands for the multiplication of Human Intelligence and Artificial Intelligence. If no human thinking is applied (HI = 0), even the best AI delivers no added value — the more competence and judgement a person brings, the more powerful the result.
Why is it important for children and students to get to know AI?
According to Thorsten Jekel, it will not be AI systems that replace people in professional life — rather, people who use AI intelligently will have a competitive advantage over those who ignore it. That is why it is already important today to introduce children and students to the competent use of AI.
Tools & Resources Mentioned
- Artificial Intelligence (AI) – Topic page with further content on AI at digital4productivity.de
- NotebookLM – Google tool for uploading learning materials and generating audio podcasts from them
- Perplexity – AI-powered search engine for source-based research
- ChatGPT – AI tool with voice mode, usable as an interactive learning and language tutor
- Duolingo – Language learning app with AI elements for pronunciation and gamified learning
