I. “Nothing but Mimicry”

A year ago I conducted an experimental dialogue with ChatGPT about Nietzsche's philosophy for this blog (link); in this I repeated the experiment (link). The result wasn't exactly exhilarating. ChatGPT continues to reproduce generalities and Wikipedia knowledge and fails when it comes to simple detailed inquiries. What it cannot do in particular, but which is the basic requirement for working in the humanities: Be good with literature. Quotes are completely misattributed and sources are invented without batting an eye that doesn't exist anyway. What Werner Herzog says about the digital experiment The Infinite Conversation, notes a fictional dialogue between himself and Slavoj Žižek conducted by AI:

I myself have a never-ending conversation with a Slovenian philosopher on the Internet, which imitates our two voices with great accuracy, but our discourse is meaningless, without new ideas, just an imitation of our voices and selected topics that we have both talked about in the past. All sentences are correct in grammar and vocabulary, but the discourse itself is dead, without a soul. He is nothing but mimicry.¹

In other words, ChatGPT and his colleagues may have a lot of chaos, but not lively, productive chaos — she could therefore not give birth to a “dancing star”, at best more or less interesting or pleasing content. How too?

‍

II. Daily, everydaily

A look at social media, where people have been virtually beaten to death with AI-generated Billo content for a few months now, is enough: Only rarely are these works really convincing; they tend to get on your nerves. Generating really good films, songs, images or texts with AI still requires creative human input; the machine alone is only able to amaze with its resemblance to such content. As a rule, after a short period of time, you recognize the specific handwriting of the AI — or, more precisely: the noticeable absence of handwriting, the sterility of its productions.

AI has also seeped into my everyday life in the meantime. At the latest after Google gives out AI answers to inquiries as standard, it is hardly possible to avoid it. Admittedly, I have sometimes had quite positive experiences with AI when it comes to everyday problems. Instead of having to laboriously click through numerous pages or even watch YouTube tutorials that last for minutes, for example, to find out at what temperature you should wash a certain fabric or how long a certain food can be kept outside the refrigerator, the AI now gives a concise and in most cases — at least for domestic use — correct answer. A few weeks ago, ChatGPT even helped me solve a pretty serious computer problem. With true angelic patience, it explained to me step by step how I could find the mistake, save my data and then fix it. And I was even able to ask curious questions about why I should take these steps in the first place and what exactly was technically behind them. The result: I saved a lot of money for a manual repair and some time scrolling through some obscure Internet forums. ChatGPT has almost done everything for me; although in the end it still needed me to implement its instructions and I probably wouldn't have been able to do that without a certain amount of previous hardware knowledge.

Of course, the qualitative leap is rather small when it comes to such banalities. Whether you collect the relevant information yourself or search for it from a computer program and have it presented in a clear way is actually of secondary importance. In any case, in my actual philosophical work, I have not yet been able to discover any real usefulness of AI.

What I also experimented with was having AI programs write entertaining stories or role-playing scenarios for me, such as “I am a pirate in the 18th century and seafaring in the Caribbean and talking to a colleague” or “I roam through the fairytale forest as a magic elf and experience funny adventures.” There, too, there was a certain astounding effect at first, until I got tired of the same redundant wording and exhausted ideas after a while and then preferred other options for evening entertainment. In order for the result to be somewhat interesting, too much personal input was simply required in the end — that makes it more exciting to read a novel or watch a movie.

‍

III. Digitalizing the Children's Room

It is sometimes more, sometimes less exciting to watch YouTube videos with my little three-year-old son in the evening. Even children's bedrooms have long since reached digitization; “watching something” has become an integral part of our everyday lives. But how do you seriously forbid the junior to enjoy yourself when you're constantly hanging on your cell phone? At least we'll make a joint experience out of it and I (still) have full control over the content. And while vacationing on the Baltic Sea, it was simply claimed that there was no Internet here — then it went without it for two weeks. The educator cannot do without noble lies, as Plato already knew.

The digitization of children's rooms is by no means a marginal phenomenon and an important aspect of the topic that is rarely talked about. Out of curiosity, I recently found out what the video portal's most successful channel actually is. In 1st place, known as unsurprising: the penetrating upper nerve saw MrBeast, whose dubious business practices I don't want to comment on here, with 423 million subscribers. In 2nd place is an Indian music channel called T-Series, which inspires 301 million followers. In 3rd place close behind with 195 million: Cocomelon — Nursery Rhymes, a channel whose content is only suitable as a torture tool for the average adult, which solely aimed at toddlers!²

When you look at the most-clicked videos, it becomes even more absurd. Here, Cocomelon is in the top 10 twice. The children's birthday hit Wheels on the Bus Is in 3rd place with 7.8 billion clicks, the Bath Song Follows closely behind with 7.12. And the most-clicked YouTube video of all time by far: Baby Shark Dance with 16.14 billion! And in the top 10, there are two more videos just for kindergarten kids!³

The recipe for success of these channels: Create veritable audiovisual drugs for the early childhood brain. These videos have no educational value, they are simply about making children addicted to these colorful, fast-changing images and catchy melodies that completely flash the child's brain. And I know for sure: As soon as I showed Junior the trash, it would be immediately attached.

But the main responsibility probably falls on the tens of millions of parents all over the world who sedate their children with such toxic content or do not have enough ego strength to simply say “no”; or whose living conditions are simply too precarious. You really have to remember it: For all ages except maybe two to six year olds, these videos are almost unbearable, which means there must be millions of children's rooms in the world in which these delusion lyrics run in endless loops.⁴ — It is hardly surprising that there have been AI offers for toddlers and parents who let their children talk to them for hours on end.

No, I don't have any of that, we still play normally and the only one who makes up stories based on “prompts” is dad. I really see little potential of AI here and would like to keep my son ignorant of the fact that it even exists for as long as possible. He should build magical worlds himself with his animal figures and blocks, invent fictional friends and paint crazy mythical creatures.

YouTube is unfortunately like a ball pit filled with sticky marzipan balls when it comes to content for toddlers. A particularly striking example: Anyone looking for “German fairy tales” or something similar must first scroll through tons of presumably exclusively AI-generated nonsense until they find something useful.⁵ So is everything lost and are our children conditioned from an early age into screen-addicted, imaginative consumer zombies? “Last person” by birth? OpenAI CEO Sam Altman recently proudly announced that children born today could never be as “smart” as AI. That may even be true if you define “smart” as mindless parroting that sounds clever.⁶ But for different reasons than Altman might think...

In any case, it seems to me that digital media should be treated with caution in early childhood education. It would be unworldly not to use them at all, but at the same time they raise children to be passive. That is a great danger.

‍

IV. The “Bird Wisdom”⁷

There is at least one island of hope: the corresponding offers from public service broadcasters. At the forefront is the classic classic since Grandpa's childhood: The Sandman. Imaginative, funny and creative stories are told here that were not simply created by an Indian content creator entering “Write me a German fairy tale with a princess, a clownfish, and a rhinoceros” on ChatGPT. And the best part: They are also entertaining, amusing and educational for parents. It's really not about parking the child in front of the screen, but cuddling together to watch something that everyone likes and that you can laugh about together and then convert.

One of our last evening stories was now particularly remarkable. It was about the young sparrow Fieps. He flies happily through the city with his sparrow parents and discovers a carding box that awakens his curiosity. It includes: a tin bird with retractable spring. Fieps wants to talk to him, but doesn't get an answer. When squeaks at him, he suddenly starts moving, yet completely clumsy and aimless. He runs into the crate wall again and again. Fieps is sorry and he wants to teach his new friend how to fly. But showing, explaining and prodding doesn't help: The tin mate just doesn't understand. Fieps then has an idea: He nudges the machine again and catapults it out of the box with a rocker made of a rolling pin and a book. The tin bird jumps from there into the evening sun, Fieps keeps his wind-up key, which he keeps from then on as a kind of cuddly toy.

It's the same with humans and AI: By itself, they can't do anything at all, you first have to pull them up and nudge them before they achieve anything. She often doesn't even understand our instructions and even though she seems human at first glance, she is simply not capable of flying fancy. You can only make them fly with a lot of your own creativity, but even then it's more of a jump towards the evening light, not a departure into dawn. And at some point it's over when there's a lack of human input. We humans have gained a new useful and entertaining toy as a result, but they are just as little threatened as chess computers or real pets from Furby (by the way, a kind of “artificial intelligence” — released 26 years ago!) — Who even knows him anymore? At the end of the day, we can go to sleep with peace of mind while the robots mop the floor and do the dishes without grumbling...

‍

V. Quid est veritas?

You always have to keep in mind that AI is not least a huge hype, a typical bubble. Large corporations have invested billions of dollars in developing these tools and they must now pay for themselves somehow. And countless fortune seekers now want to have their slice of marzipan cake off. From this point of view must The AI will be amazing, revolutionary, incredibly labor-saving. And in part, she certainly is all of that. But in the end, we are dealing with a phenomenon such as cryptocurrencies, automatic driving or the miracle toy mentioned above: topics that go through all media for a few months and yet in the end turn out to be air numbers or at least not as life-changing as they are initially presented.

I feel a certain calmness from Nietzsche when it comes to such short-lived questions, the “noise of great actors and the whirring of poisonous flies.”⁸. The main aim is not to succumb to a mistake of thinking that is often made in the current media debate about AI: to assume that technological progress is linear, if not exponential. The truth is: New technological innovations usually appear suddenly, but their development then often stagnates over a longer period of time and only concerns detailed questions until the next technical revolution takes place. Technological development is therefore sudden and sudden, not continuous and predictable. In the case of AI, it will now primarily be a matter of exploring the limits and possibilities of its everyday use by means of better interfaces; further groundbreaking developments are not expected in the foreseeable future, if only for the reason that the hardware will set narrow limits to its linear development. Computing power is physically limited as long as completely new chip models are not invented, and even the power that powers the supercomputers behind the AI does not simply come out of a power outlet. Just as robots will not be able to deliver packages or provide supermarket shelves in the foreseeable future, they will be able to overcome the fundamental problem that human creativity and genuine thinking include “chaotic” elements bound to a sentient and suffering body that computers can only imitate but cannot do themselves. Nietzsche also clearly recognized this.

Let's not kid ourselves: There is actually no “artificial intelligence,” but only a significant improvement in the language and image and sound generating capabilities of the machines we have developed. Computers have gotten better at deceiving us and simulating intelligence; they haven't really become intelligent. This is most obvious when it comes to the question of the truth of statements. We humans too spend most of the time just parroting on statements that we have picked up somewhere, but at the same time we have an idea of the truth or falsehood of these statements. On the basis of a specific intuitive understanding of our reality, we make judgments about the truthfulness of statements — every day. But this process is a highly creative one, which requires interpreting a statement in such a way that it can be translated into truth criteria and to verify their fulfillment on the basis of, in the last instance, a bodily and intuitive understanding of reality.

Even a computer that was fed with a database that contained all the facts in the world — which is impossible in itself, since even the simplest phenomenon can be broken down into an infinite number of “facts,” if there is no intuitive preconception of what should and should not be considered a relevant fact — could impossible decide on this basis whether the statements it produced are true or false because it completely lacks such an understanding. The AI's tendency to hallucinate is therefore not an “accident” or a technical shortcoming, but reveals an insoluble aporia: There can be no machine that can distinguish between truth and falsehood without precisely specified parameters. But since there are usually no such clear and unambiguous parameters for ordinary statements, it has to be just parroting. All you have to do is construct sufficiently complex cases — such as the interpretation of philosophical texts — to make every AI fool; and the most mundane questions are often enough.

And these considerations apply analogously to all areas of human practice. The successful conditions for our activities can never be 100% operationalized in a machine-compatible manner. Characteristics of a good meal, a polite gesture, a virtuous act, a beautiful landscape; we all know them without being able to explain them — and yet it is always possible for a good cook to surprise us with a meal that we enjoy without it being in line with our explicit understanding of a good meal. And who would want to live in a world without “chaos,” in which things would be different, without exceptions, without complex borderline cases, without discretion?

‍

VI. Midwife, Nothing More

From AI itself, it is neither expected to realize any utopias nor to fear the apocalypse. After the artificial hype subsides, their limits will quickly become apparent and the discourse around them will normalize. The two really decisive questions that need to be clarified will then hopefully become the focus of attention: How can you ensure that not only large corporations but also creative data providers (artists, authors, musicians, etc.) benefit from the commercial use of AI systems? It seems to me that there is actually a major problem here in terms of genuine revolutionary expropriation.⁹ And how can we satisfy the rapidly increasing cravings for energy and other limited resources such as cooling water or metals of all kinds as a result of these new machines without further accelerating the impending ecological collapse of the planet?

Based on my cautiously optimistic assessment a year ago (link) I would therefore like to keep it unchanged. By itself, AI will not lead to an “end of the human being” — at most in the sense that a Third World War led by fully automated killer drones that can no longer be stopped breaks out, but then it is once again not the machines that are to blame — nor to the advent of the “superman.” Either way, we will have to continue to rely on our necessary limited intelligence — groping, experimenting, chaotic — since we are not “thinking, writing and speaking machines.”¹⁰. Thankfully!

‍

Literature

Herzog, Werner: The future of truth. Munich 2024.

‍

Footnotes

1: The future of truth, p. 10. The book generally contains some interesting insights about AI.

2: Source: wikipedia, as of 19.08.2025.

3: Source: wikipedia, as of 19.08.2025.

4: You can also calculate it more precisely. The Baby Shark Dance has been online for 9 years. There are around 1.45 billion children in the world between the ages of 0 and 10, of whom around 1/3 live in extreme poverty (!) and presumably has no access to such media, leaving 957 million. This means that every child in this cohort must Baby Shark Dance Have viewed an average of 17 times since it was published!

5: My recommendation at this point: The unfortunately discontinued channel Your fairytale world.

6: Which, by the way, is also quite similar to Altman's understanding of Parvenues.

7: Thus Spoke Zarathustra, The Seven Seals, 7.

8: Thus Spoke Zarathustra, The flies of the market.

9: However, this question is also likely to prove to be solvable, as creative companies such as Disney are now also taking on the matter (link).

10: The benefits and disadvantages of history for life, paragraph 5.