I fell into it myself.
You read something and hesitate. Is this a human speaking? Or a system optimizing language?
I’ve written texts myself that were coherent and convincing, yet something of my own voice had disappeared. Since becoming more aware of it, I notice that same smooth uniformity everywhere.
A text can flow beautifully and still become flatter. More general. Less unique. Less embodied. The personal story grows thinner, nuance fades, sentences become predictable. That’s what makes it uncomfortable. Not just stylistically, but at a deeper level: are these words still connected to direct experience, or are they mainly moving on the level of abstraction?
I recognize that tension because I regularly work with AI, both in my writing and in my practice. My work constantly moves between direct experience and abstract thinking - between feeling what is present in the moment and organizing it into patterns. AI dramatically amplifies that abstract layer. For me, the question is not whether that is good or bad, but whether our words remain rooted in our own experience.
AI optimizes for probability: what best fits what is already there. But what statistically fits a context is not automatically what feels most true to me - or most my own. That’s where the friction lies.
If AI is unavoidable, how do we work with it without losing ourselves? How do we make sure that what we write still carries our own voice?
P.S. This piece is a bit longer than usual. That’s intentional. Complex questions rarely fit into a few paragraphs without simplifying reality.
This Didn’t Start with AI
What we are seeing now with AI is not entirely new. It accelerates something that has been unfolding for much longer.
Since the invention of the printing press, we have increasingly captured experience in transferable forms. Writing made knowledge reproducible. Ideas could travel. Generations could build upon one another. It was revolutionary.
But something subtle also happened. What was once an encounter - a voice in the room, a body across from you, a moment unfolding in relationship - became something that could be recorded and detached from its original context. A text. A document. A report.
What had taken place between people became a representation of what had happened. Distance emerged between experience and description.
Television accelerated that process. Images replaced presence. Social media compressed it further. Complex stories became fragments. Nuance gave way to speed.
In an earlier blog, I wrote about an edited newspaper article based on a podcast interview. In the podcast, you could still feel the speaker’s involvement - the hesitation, the urgency, the conviction. In the written version, mostly the information remained. It was factually correct, yet something essential had vanished.
This is not an AI problem alone. It shows how easily lived experience shifts into representation, and representation into model. How something once alive becomes transferable and optimizable.
On Abstraction and Distance
Abstraction is a tremendous human capacity. Without abstract thinking there would be no legal systems, no economies, no technology - and no AI. It allows us to recognize patterns larger than our individual perspective.
But abstraction also creates distance.
Anthropologists describe how some African communities that had not learned to read and write approached logic and generalization differently. Thinking remained closely tied to concrete experience and specific situations. Once writing is introduced, reality becomes movable and systematizable. New possibilities emerge.
And so does distance from direct experience.
That distance is necessary to operate at scale. But the greater the distance, the easier it becomes to speak without seeing the face of the person you are speaking about.
That is where dehumanization begins. Not with hatred, but with abstraction disconnected from lived reality.
What Becomes Visible at Scale
The same dynamic appears in organizations and institutions. As soon as policy becomes detached from practice, distance grows. At the policy level, we work with models, criteria, averages. That is necessary to function beyond the human scale.
But the greater the distance, the easier it is for the individual story to disappear.
The Dutch childcare benefits scandal offers a painful example. What began as systemic logic and risk assessment ended in human devastation. The issue was not abstraction itself, but abstraction without feedback from lived reality.
When distance becomes structural, simplification often follows. Differences turn into oppositions. The other becomes a label instead of a person.
This mechanism is not only political but psychological. The more we become separated from direct experience - from our own bodies, from nuance - the more appealing black-and-white thinking becomes. In many traditions, this is described as living from separation or duality. But it often begins in something very concrete: losing contact with what we actually experience ourselves.
Technology such as AI operates precisely on that abstract level. That makes it powerful. And it requires us to actively bring lived experience back into the equation.
Should I Avoid AI Then?
If abstraction so easily detaches from experience, the question naturally arises: should I simply not use AI?
That would be too simple.
We asked similar questions about books, about computers, about the internet. Every new layer of abstraction expands our possibilities and our distance at the same time. The problem is not the tool itself, but how we relate to it.
Avoiding AI altogether would mean placing oneself outside the society we are part of. The question is not whether we use it, but how.
AI offers real advantages. It analyzes large amounts of information quickly, reveals patterns, helps structure and organize, corrects language, automates routine tasks, and increases capacity without requiring linearly more time.
I notice that daily in my work. During a session, I no longer need to take extensive notes and can remain fully present in the interaction. Analysis that would otherwise remain scattered across notes can be reviewed coherently afterward.
But efficiency is not the same as depth.
Individually, AI can make us more productive. At the same time, it can subtly shift how we think and write. The convenience and optimization of AI quietly start shaping how we formulate and reason.
That raises another question: how do we use these tools without them gradually shaping the way we think?
And in that same process something else becomes visible as well: what AI cannot replace - the direct, relational attunement that happens in the moment itself.
How I Use AI in Writing — and Where It Frictions
When I started a consulting firm in 1999 with a journalist, we launched something innovative at the time: a newsletter with a provocative column. I brought in ideas and vision; my colleague acted as sparring partner and editor. Later, others in my network took on that role.
Much of what they did then, AI can now do in many respects.
AI helps me organize my thoughts. It detects inconsistencies. It suggests sharper phrasing. It brings structure where my writing is still searching. It is fast, analytical, tireless.
But writing is not a repeatable protocol. It is an associative process in which each text emerges anew. AI can support and refine it, but it cannot replace it. And it is precisely in that support that friction arises.
Almost every time, the same thing happens. A passage becomes more compact. An example disappears. A sharp formulation becomes more neutral. A sentence that still carried tension gets neatly resolved. The whole becomes more coherent.
And then I read it back and think: something is gone. Not because it is wrong, but because it has become flatter. I respond with things like:
Why did you remove that example?
There was still nuance there; it should remain.
The sentence is technically correct, but it no longer feels like my language.
You are building on one paragraph, but not on the text as a whole.
These are small adjustments. Yet they point to the same principle: no reduction without necessity. No abstraction without an anchor. No smoothing out what still carries friction. No sentences that do not feel like mine.
AI optimizes for coherence and probability. That is its strength. But probability is not the same as individuality. What statistically fits a context is not automatically what feels alive. That is the tension.
If I am not attentive, the text becomes clearer and neater, and at the same time less unique. What was layered becomes linear. What carried multiple meanings becomes singular.
Feedback as Part of the Process
That is why I give AI feedback. I reinstate passages. I ask to reintroduce a layer that disappeared. I correct when something has been compressed too much. Sometimes I simply say: this is correct, but it does not land.
I go further. I regularly ask: what have you learned here? What patterns do you see? What should you remember for future texts? In that way, it becomes an ongoing learning process - not only for me, but for the system I am working with.
The more precise I am about what does and does not fit, the better the system follows my tone and way of thinking. Responses become less generic and more attuned.
That requires attention and time. Without that feedback, flattening happens naturally.
Not because I formulate one perfect prompt, but because I keep responding to what is actually happening in the text and make explicit what needs to remain.
How I Use AI in My Sessions
One advantage of working online is that I can record sessions with Zoom. That supports clients in integration. As I often say, it gives them something like a free additional session.
Zoom automatically generates transcripts when recorded in the cloud. With the client’s consent, I ask AI to summarize the transcript - not only in terms of content, but also process: what happened in the dynamic? Which interventions were supportive? Where did something shift?
Clients receive a structured summary where both content and process are visible. And I can quickly revisit what truly unfolded in the previous session.
In the beginning, those summaries were often too smooth and too general. AI optimized for clarity and compactness, but in doing so lost richness. Subtle interventions - the moments where something turns - disappeared into generic phrasing. The text was correct, but the lived experience had been filtered out.
There was no strict instruction protocol. I did not create a checklist from the start. It was situational. I read a summary and sensed: something is off. Sometimes it was one sentence. Sometimes the tone. Sometimes what was missing. I gave feedback. The response adjusted. Then the next refinement followed.
It was an iterative, relational process. Only afterward could I abstract it into guidelines:
Work through the entire transcript before summarizing.
Keep both content and process visible.
Name interventions concretely.
Add no assumptions.
Do not mix sessions.
Only refine style after completeness.
But those were not starting instructions. They are the abstraction of a process that emerged in real time.
When I recognized that, I saw the same mechanism at another level.
What This Taught Me About Learning
Learning rarely begins with a fully developed protocol.
First there is experience. Then friction. Then adjustment. Only afterward does a pattern become visible.
That is how people learn.
That is how I help clients recognize their own mechanisms.
And that is how I gradually taught AI to integrate my way of seeing. Not by writing one clever prompt, but by repeatedly indicating what fits and what does not. By guarding nuance. By continuing to feel the difference between a text that is technically correct and one that feels true in experience.
Rules come afterward. They are the residue of attention.
This goes deeper than prompting. A prompt is an instruction. What happened here was something else: ongoing refinement based on lived experience. Not optimizing output, but adjusting according to what I sensed.
AI supports recording and structuring. It can organize what was said, reveal patterns, distinguish content from process. But what happens in the encounter itself - the quickening breath, the silence that falls, the moment someone dares to feel something fully - arises only in presence.
You can describe it afterward. You can analyze it. But it always begins in direct experience.
Reconnecting Abstraction to Experience
Clients can often explain their patterns flawlessly. They understand the theory. They can name the label. Yet their breath quickens when a certain topic arises. Their shoulders tense. Their voice shifts.
That is where I begin. Not with a model. Not with analysis. But with what is present now.
We slow down. Attention returns to direct experience. What is happening in the body? Where is tension? What is being avoided?
At some point vulnerability appears. Someone not only notices tension, but allows what lies beneath - insecurity, shame, fear of rejection. That is the turning point. Not the insight itself, but allowing what is actually felt.
Only then do we abstract. We explore the mechanism. What triggered it? Which pattern took over? What older meaning is underneath?
Then abstraction becomes integration rather than distance.
Understanding only gains power when it remains connected to experience. Otherwise it stays a mental insight without shift.
From that embodied place, I also look at AI.
Staying Rooted with AI
What becomes visible to me here is not only about technology.
It is about small shifts. A text that sounds smooth but does not quite ring true. A sentence that is logical but no longer yours. The moment you stop rereading to check whether it still genuinely fits.
Development works organically. Major changes arise from repeated small shifts. Trauma often forms not in one dramatic event, but through accumulation of moments where something was not met.
The difference lies in how we respond.
The same applies to how we use AI. Whether we continue to sense when something is still ours - in what we say and in how we say it.
Abstraction and experience need each other. When thinking detaches from feeling, distance emerges. Without structure, experience remains diffuse. Development unfolds in the movement between them.
AI does not change that movement. It amplifies it.
The question is not whether we use technology, but whether we repeatedly return to direct experience in what we do.
And that is precisely where my work lies.
