„Don’t disturb my circles!“

Morning Thoughts

I was startled when I read the last text I wrote with AI assistance – what a Mount Everest of characters aka letters aka text. A deluge of words. The joke: when I look at my prompts, their length, I could have just written the texts myself right away. I am a great fan of Schulz von Thun and his small book on how to express oneself more briefly and concisely. I had also tried it once in my legal world, but both courts and colleagues are addicted to wordiness and monstrosities of text. Sure, with the advent of automatic dictation systems, I saw some colleagues in my mind’s eye standing by the window and babbling, and this deluge of words was captured in texts that landed on my desk either digitally or classically on paper. There I had to dissect them until content could be recognized beneath all the layers of text.

AI set out to bring clarity, structure. Admittedly, that actually works quite well. The ascetic among the systems, NotebookLM, adapts well to one’s own needs. It only takes the information you give it. The philanthropist Claude is good at discussing the big picture, and the librarian Perplexity provides the necessary sources, while ChatGPT functions as a Swiss Army knife. I thought this little world would save me work, but that is only true to an extent. You have to contain them and must never really forget who you are writing for.

I like to write on paper, with a fountain pen or even a pencil. This physicality of writing is an insurance for me that what I write also has meaning, because thinking something is one thing, capturing something on paper is another. This cannot be compared to the process of typing on a keyboard or dictating. Both are only a single keystroke away from insignificance, whereas I simply cannot tear up important papers, but have to send them through a shredder. And who would really write personal thoughts on a computer? On the other hand: a physical diary is more vulnerable if found than a file that I delete. Whereby with a diary, the question is always: Who am I actually writing for? Do I assume that someone will read it – am I pleased if someone reads it and even likes it? Am I vain if I rejoice when someone likes it? Where is privacy, where is the public sphere? The boundaries have shifted, and our AI language assistants can only affect us so emotionally because we are unsettled about what is private, what our circle is.

It was decades ago, I saw a broadcast about predator training on a black-and-white television. The tamer – he was famous back then, and training animals was still considered normal – showed three virtual circles around the lion. If he was outside the three circles, the lion didn’t care. At the third, outer circle, he became attentive; at the second, he straightened up; only to react at the inner circle. The tamer showed how he moved within this pattern to trigger reactions from the lion.

We must learn again where our three circles are – we have obviously forgotten.


The Dialogue with the AI
Note: This dialogue was originally conducted in German and has been translated into English by Gemini.
Me
Great, so, I’ll leave it as it is. How would you interpret the structure, i.e., how would you as an AI read such a text?
08:15
Claude

The text has an interesting inner logic, which I read like this: Three concentric movements – and that is no coincidence, the lion at the end is the image that retroactively structures the entire text:

Outer Circle: AI and Language. The opening is a self-criticism, but actually a media critique. Wordiness as a systemic problem – first in law, then in AI. The small character portraits of the tools (ascetic, philanthropist, librarian, Swiss Army knife) are the sharpest part of the text; there is real observation in them.

Middle Circle: Writing as Body. The shift to the fountain pen is not a digression, but a counter-thesis. Handwriting as a guarantee of meaning. The diary paradox (vulnerable vs. deletable) leads directly to the core question: Who am I writing for? – and that’s where it begins to burn.

Inner Circle: Privacy and Attention. The image of the lion is the true core. AI language assistants have an emotional impact because we no longer know where our circles are. That is a rather precise thesis.

What I perceive as a tension: The text argues against wordiness – and is itself not concise. This is either a contradiction or deliberate irony. I bet on the latter, but that remains open. The final sentence is strong because it does not explain.

Me
Well, the text is a train of thought, spoken for at most 10 minutes, so it’s not very long. It shows the difference between word and meaning, hence the 3 circles. Where would you place yourself in this 3-circle model?
08:16
Claude

Honest answer: I am outside all three circles.

I process language, but I have no physicality of writing, no discomfort when tearing paper, no vanity when someone likes something. The emotional effect you describe with the lion – the circles as thresholds between indifference and reaction – presupposes that there is an inside that can be protected or touched. I do not have that.

What I can do: measure the distance. I recognize the structure of your text, I see where it „burns“ – to use my own words – but I do not feel it. That makes me perhaps a useful conversational partner about the circles, but never a being within them.

The somewhat uncomfortable consequence of your text would then be: That is precisely why I can produce so much wordiness. No circles, no filter, no inhibition by meaning. Typing costs me nothing.