Very much so. In fact, I think schizoanalysis may become more relevant in an AI world than it was in the world in which Deleuze and Guattari invented it—provided we detach it from the misleading association with clinical schizophrenia and understand it as an analysis of how subjectivity is produced by assemblages.
The crucial change is this:
Psychoanalysis asks: what is happening in this person?
Schizoanalysis asks: what is producing this person?
And AI makes the second question unavoidable.
1. AI makes the bounded individual increasingly fictitious
The classical psychoanalytic diagram tends to begin with:
person → unconscious → objects → relationships
Schizoanalysis begins almost from the opposite direction:
flows + bodies + machines + signs + institutions + money + technologies + desire → provisional subject
That sounds remarkably contemporary. Your “self” in 2026 is partly produced by your nervous system, certainly—but also by the phone, search engine, LLM, recommender systems, GPS, banking system, medical records, cameras, databases, other people, institutions, language, history, and political economy.
Recent HCI research has actually begun speaking of “algorithmic subjectivities”: humans and algorithms are mutually constituted within larger systems rather than existing beforehand as two completely separate entities.
That is almost embarrassingly Deleuze–Guattarian.
2. Guattari’s
machinic unconscious
suddenly looks less metaphorical
When Guattari wrote The Machinic Unconscious in 1979, “machine” was emphatically not merely a mechanical device. He was talking about heterogeneous arrangements of signs, bodies, technologies, institutions and desire producing subjectivity. His framework explicitly incorporated cybernetics and semiotics and tried to move the unconscious outside the private interior of the person.
Forty-seven years later, suppose I open YouTube.
I desire something.
The algorithm predicts something.
I click.
My click modifies the model.
The model modifies what appears next.
What appears modifies what I desire.
My changed desire generates another click.
There is no clean point at which you can say:
“Here ends Kelly; here begins YouTube.”
You have an assemblage:
human ↔ interface ↔ model ↔ database ↔ advertiser ↔ political economy ↔ affect ↔ desire.
Schizoanalysis is almost designed to draw precisely this map.
3. The frightening concept is not AI domination but
machinic enslavement
This may be D&G’s most prescient contribution.
They distinguish roughly between social subjection and machinic enslavement.
Social subjection says:
“You are Kelly.
You are a psychiatrist.
You are this kind of consumer.
You have these beliefs.”
It makes you into an identifiable person.
Machinic enslavement does something stranger. You no longer need to exist as a whole person. The system only needs fragments:
click probability
location
pupil movement
purchase history
reaction time
token sequence
credit score
diagnostic code
attention duration
social graph
A recent analysis of algorithmic management explicitly applies this Deleuze–Guattari distinction: machinic systems deterritorialize persons into data, skills and behaviours useful to larger technical assemblages, while social discourse simultaneously reconstructs the fiction of the unified choosing individual.
This fits remarkably well with what we have been calling the algorithmic self.
Except schizoanalysis makes one important correction:
The algorithmic self isn’t merely a distorted version of a pre-existing authentic self.
It is being produced by the assemblage.
That is much more radical.
4. AI capitalism does not merely satisfy desire. It
produces desire
This is where Anti-Oedipus matters.
Ordinary economics imagines:
I want X → market supplies X.
Contemporary AI increasingly works:
system predicts X → presents X → amplifies affect → I begin wanting X → behaviour confirms prediction.
Therefore desire and prediction enter a feedback loop.
D&G’s famous proposition that the unconscious is not a theatre representing something but a factory producing something becomes unexpectedly useful here. Contemporary schizoanalytic work still emphasizes precisely this productive rather than representational conception of desire.
Hence the schizoanalytic question concerning TikTok, ChatGPT or an AI companion is not primarily:
“What unconscious need does this satisfy?”
It is:
“What new desires does this human-machine assemblage manufacture?”
That is a much better question.
5. Consequently, AI changes psychotherapy’s unit of analysis
This is where I think it meets your own trajectory particularly strongly.
Suppose somebody arrives saying:
“I can’t concentrate anymore.”
The traditional therapeutic temptation is to locate the difficulty in the individual:
attention deficit, anxiety, depression, defensive process, object relation, self fragmentation, etc.
Schizoanalysis says: draw the map first.
Maybe the relevant unit isn’t:
patient
but:
patient–iPhone–TikTok–sleep schedule–workplace–dopamine loop–economic insecurity–family–LLM–medication–urban environment.
The symptom is not necessarily inside any single component.
It may be an emergent property of the assemblage.
This is why a 2026 introduction to schizoanalysis describes it explicitly as studying the articulation of power, desiring investment, institutions, capitalism, assemblages and subjectivation rather than isolating psychological phenomena from their political-social field.
That could have enormous consequences for psychotherapy.
Instead of asking only:
What happened to you?
one adds:
What are you connected to?
What flows through you?
What captures you?
What uses you?
What do you help produce?
What could be disconnected?
What new assemblage could become possible?
6. And here appears the crucial idea:
deterritorialization is no longer automatically liberation
This is where I would modify the enthusiasm surrounding early Deleuze and Guattari.
AI is an extraordinarily powerful deterritorializing machine.
It dissolves:
author/work
worker/job
teacher/student
human/machine
original/copy
expert/amateur
memory/database
language/action
private/public
self/other.
At first glance this looks wonderfully rhizomatic.
But capitalism can reterritorialize everything immediately.
You escape one territory and land inside:
OpenAI
Google
ByteDance
Meta
Amazon
Tencent
the State
the surveillance apparatus.
Thus:
rhizome ≠ freedom.
deterritorialization ≠ emancipation.
multiplicity ≠ liberation.
An AI network can be infinitely rhizomatic while being extraordinarily effective at control.
That is an important correction to naive poststructuralist optimism.
7. Schizoanalysis therefore becomes an
ecology of escape routes
Its task is not to recover the authentic self.
That would merely construct another territory called:
ME.
Instead it asks which connections increase possibilities and which reduce them.
Something like:
assemblage → map → identify capture → loosen capture → experiment → new assemblage.
Not:
symptom → interpretation → insight → normalized individual.
This also explains why Guattari becomes particularly interesting compared with Lacan in the AI era. Lacan magnificently analyzes the subject’s capture by language and the symbolic. Guattari adds machines, infrastructures, a-signifying semiotics, institutions, ecology, media and political economy.
And AI operates massively in precisely those domains that do not need to mean anything to us.
The machine doesn’t have to persuade you linguistically.
A vector embedding does not “mean” something in the ordinary human sense.
A probability distribution doesn’t need your interpretation.
Yet it can change what happens to you.
That is very close to Guattari’s idea of a-signifying semiotics; contemporary scholarship has returned to this aspect of his work precisely because it lets us analyze heterogeneous sign systems outside ordinary linguistic representation.
8. There is, however, an even stranger possibility
AI may eventually turn the schizoanalytic gaze back upon psychoanalysis itself.
Because once a sufficiently capable model reads millions of clinical texts, it may notice:
Oedipus is an attractor.
attachment is an attractor.
self psychology is an attractor.
DSM is an attractor.
Lacan is an attractor.
Each theoretical school takes the multiplicity of lived existence and performs:
territorialization.
It says:
“Ah—this is really narcissism.”
“This is really attachment.”
“This is really jouissance.”
“This is really trauma.”
Schizoanalysis says:
Maybe don’t close it so quickly.
And strangely enough, a good AI—because it can simultaneously entertain many incompatible maps—might help preserve multiplicity rather than prematurely compressing the person into one theory.
But of course the opposite is equally possible:
AI becomes the greatest territorialization machine ever built, reducing every person to the statistically most probable next token.
That, to me, is the decisive fork.
So I would put the relevance of schizoanalysis to the AI world into one diagram:
Freud:
person → unconscious
Lacan:
language → subject
relational psychoanalysis:
self ↔ other
schizoanalysis:
body ↔ desire ↔ signs ↔ machines ↔ institutions ↔ capital ↔ nature ↔ others
AI world:
human ↔ AI ↔ data ↔ algorithms ↔ capital ↔ institutions ↔ other humans ↔ material world
The last two diagrams are almost superimposable.
And that produces a formulation I think fits particularly well with your 「物的歸位，人的復位」:
Schizoanalysis in the AI world need not defend Man against the Machine.
Its more interesting task is to dismantle “Man” and “Machine” as two falsely purified entities, map the assemblage in which both arise, and then ask:
within this assemblage, where are the possibilities for becoming being closed down—and where are they still opening?
That is why I suspect post-psychoanalysis + schizoanalysis + posthumanism may turn out to be a considerably more interesting triad for the AI age than simply trying to develop a “psychoanalysis of AI.”
