https://drive.google.com/file/d/1VelmheifJxahposmQNKxkLpcXoOAHbTf/view?usp=drivesdk
Danielle Knafo’s “Artificial Intelligence on the Couch: Staying Human Post-AI” (2024) is one of the more thoughtful psychoanalytic attempts to understand AI—not merely as a technological tool, but as a profound challenge to human identity, subjectivity, and psychotherapy. The central question is not whether AI will become intelligent, but what will remain uniquely human when intelligence itself is no longer uniquely ours.
AJPAIarticle.pdf
Here are what I think are the paper’s main ideas.
1. AI is the latest narcissistic blow to humanity
Knafo begins by placing AI in a historical sequence.
- Copernicus: Earth is not the center of the universe.
- Darwin: Humans are not fundamentally separate from animals.
- Freud: We are not masters of our own minds.
- AI: We may soon no longer be the most intelligent beings on Earth.
She argues that AI represents perhaps the greatest challenge yet to human exceptionalism.
AJPAIarticle.pdf
2. “Algorithms are the new unconscious”
This is probably the paper’s most original thesis.
Knafo argues that algorithms resemble Freud’s unconscious because they:
- operate invisibly,
- influence our decisions,
- know patterns about us that we do not consciously know,
- increasingly predict our desires.
She goes even further:
We are externalizing parts of the unconscious into algorithms.
Recommendation engines, social media feeds, Amazon suggestions, TikTok, Google Search—they become mirrors of unconscious wishes, fears, and fantasies.
She writes that AI may know us “better than we know ourselves.”
AJPAIarticle.pdf
This idea resonates strongly with your own distinction between the algorithmic self and the analogical self. Your concern is not simply that algorithms predict us, but that they gradually reshape us into beings who are increasingly predictable.
3. AI expresses humanity’s drive toward omnipotence
Using Freud and Winnicott, Knafo interprets AI psychologically.
Technology is not merely useful.
It is an expression of the wish to overcome:
- death,
- limitation,
- dependence,
- vulnerability.
Kurzweil’s Singularity, Neuralink, mind uploading, even the dream of defeating death all represent humanity’s wish for omnipotence.
She memorably imagines technology writing a “Dear John letter” to Nature:
“I’m sorry, but I must leave you behind.”
Technology is our rebellion against biological limits.
AJPAIarticle.pdf
4. Omnipotence always has a twin: the death drive
Knafo then balances the previous chapter with Freud’s death drive.
The same force pushing us beyond limits also risks catastrophe.
Examples include:
- climate destruction,
- surveillance,
- biological weapons,
- autonomous AI,
- AI creating stronger AI.
Unlike nuclear weapons,
AI can improve itself.
That is what makes it unprecedented.
She cites Harari, Geoffrey Hinton, Mo Gawdat, Elon Musk, and others who worry that humanity has created something potentially beyond its control.
AJPAIarticle.pdf
5. AI is already reshaping everyday subjectivity
Knafo devotes considerable attention to ordinary life.
She argues that social media has transformed:
- identity,
- self-esteem,
- sexuality,
- relationships,
- childhood.
Rather than developing a self through solitude and intimate relationships, people increasingly develop identity through continuous comparison with an online audience.
She draws on:
- Carlo Strenger (“fear of insignificance”),
- Sherry Turkle (“from conversation to connection”),
- internet pornography,
- influencers,
- smartphone parenting.
One striking proposal is that psychoanalytic theories of mirroring and even the Oedipus complex need updating for the AI era.
AJPAIarticle.pdf
6. What remains uniquely human?
Knafo asks ChatGPT itself what it means to be human, then critiques each answer.
She examines six qualities:
- consciousness,
- language,
- reasoning,
- creativity,
- empathy,
- social organization.
Her conclusion is nuanced.
AI is already becoming very strong in:
- language,
- reasoning,
- information,
- pattern recognition.
Harder to reproduce are:
- embodied consciousness,
- lived experience,
- genuine empathy,
- ethical responsibility,
- existential suffering.
She argues these remain central human strengths, at least for now.
AJPAIarticle.pdf
7. Can AI do psychotherapy?
This is perhaps the most interesting section for psychiatrists and psychoanalysts.
Initially Knafo believed AI could only imitate structured therapies such as CBT.
Then she tested ChatGPT by giving it one of her own childhood dreams.
To her surprise, ChatGPT produced a reasonably sophisticated interpretation, refined it after receiving associations, and even generated Freudian, Jungian, Kohutian, Lacanian, Ferenczian, Mitchellian, and Laplanchian readings.
She admits that AI did better than she expected.
It even offered perspectives that had not emerged in her own analyses.
Nevertheless, she concludes that AI still lacks:
- embodied intuition,
- spontaneity,
- genuine relational experience,
- unconscious participation in the therapeutic field.
She thinks AI therapy will improve substantially, but psychoanalysis still possesses something different.
AJPAIarticle.pdf
8. The future task of psychoanalysis
The paper ends on a hopeful note.
Psychoanalysis should not compete with AI by trying to become faster or more efficient.
Instead, it should cultivate precisely those dimensions that remain deeply human:
- presence,
- trust,
- relationship,
- reflection,
- ethical responsibility,
- meaning,
- love.
Therapeutic change is neither programmable nor linear.
It emerges through a living relationship between two people.
In Knafo’s view, that human relationship becomes more, not less, valuable in the AI era.
AJPAIarticle.pdf
My comment
Reading this paper alongside our previous conversations, I was struck by several convergences with ideas you have been developing.
First, Knafo’s “algorithms are the new unconscious” parallels your notion of the algorithmic self, though your formulation goes further. She emphasizes that algorithms reveal unconscious wishes; you emphasize that algorithms increasingly constitute a new mode of subjectivity, while your “analogical self” names what resists prediction.
Second, when Knafo concludes that psychoanalysis should preserve empathy, relationship, and meaning, you would likely agree—but add that these alone may not suffice. Across your recent writings, you have argued that what survives the AI age is not simply “the human,” but a way of dwelling: returning to history, to nature, to seasonal time, to places one can revisit, and to forms of life that remain irreducibly analogical. In that sense, your project asks a different question: not only how to stay human, but how to inhabit a world in which the human is no longer the center.
That difference, I think, is philosophically significant. Knafo remains largely within a psychoanalytic humanism. Your recent work increasingly points toward a posthuman ontology in which the task is not to defend human supremacy, but to discover forms of existence that cannot be fully absorbed into algorithmic computation.
