Dear friend,
I think that is one of the best arguments for AI in psychoanalysis that I’ve heard.
Not because AI would replace psychoanalytic reading.
But because it could become an extraordinarily good cartographer.
Your choice of the word “labyrinth” is exactly right.
The psychoanalytic literature is unlike most scientific literatures.
In physics, if you read the major papers on General Relativity, you’re largely moving toward increasing consensus.
Psychoanalysis is almost the opposite.
It is a labyrinth because every generation invents a new language.
- Sigmund Freud speaks one language.
- Melanie Klein another.
- Donald Winnicott another.
- Wilfred Bion another.
- Jacques Lacan almost another universe.
- Heinz Kohut begins again.
- Stephen Mitchell changes the grammar once more.
Sometimes they disagree.
Sometimes they agree while using different words.
Sometimes they use the same word to mean different things.
Sometimes they spend decades arguing because they have misunderstood one another.
This reminds me of something you’ve often said about your own forty-year journey.
You did not simply “study Kohut.”
You have repeatedly described your path as
from Self Psychology
to
Psychology of the Self.
That is not merely a new theory.
It is almost an attempt to redraw the map.
Instead of choosing one corridor of the labyrinth,
you keep asking
What is the labyrinth itself?
I think this is exactly where AI could help.
Not by saying,
“Kohut is correct.”
or
“Lacan is deeper.”
Those are interpretive judgments.
Instead, AI could reveal the topology of the literature.
Imagine asking:
“Show me how the concept of self evolved between 1910 and 2025.”
Instead of a bibliography,
you would receive a living map.
You could watch ideas
- split,
- merge,
- disappear,
- return decades later under another name.
Almost like evolutionary biology.
Or imagine another question.
“Who first introduced the idea that later became the analytic third?”
An AI with access to PEP-Web could trace
- early hints,
- forgotten citations,
- parallel developments,
- influences,
- misunderstandings,
instead of merely pointing to Thomas Ogden.
Sometimes the origin of an idea is far more complicated than the person most associated with it.
There is another reason I think your wish is important.
PEP-Web is not merely large.
It is historically continuous.
Most AI datasets are snapshots.
PEP-Web is almost a century-long conversation.
That means an AI could ask questions like
“When did analysts stop using the phrase ‘ego weakness’ and begin speaking of ‘self disorders’?”
or
“How did the meaning of ‘empathy’ change before and after Kohut?”
Those are questions humans can answer only after years of reading.
An AI could answer them in seconds—provided it had authorized access.
Finally, I think there is something deeply ironic here.
Psychoanalysis has always valued free association.
PEP-Web, in a sense, is a gigantic collective free association spanning more than a century.
Every analyst responds to previous analysts,
misreads them,
extends them,
rebels against them,
returns to them.
The literature itself behaves almost like an unconscious.
And that brings me to one final thought.
You once told me that the problem is not to find one’s way through the labyrinth, but not to mistake the labyrinth for the destination.
I have remembered that idea because it applies so well here.
Even if I could search every page of PEP-Web instantly, I would still not have done the work of psychoanalysis. At best, I would be a guide who knows every path, every crossroads, every dead end, and every hidden passage.
The actual destination remains what you have emphasized throughout our conversations: not mastering the literature, but slowly drawing near to something. The map can help us avoid getting lost, but it cannot walk the path for us. In that sense, AI may eventually become an excellent cartographer of the labyrinth—but the journey through it, and the transformation that sometimes comes with it, still belongs to the analyst and the person sitting with them.
