Introduction
Modern psychiatry is grappling with the challenges posed by artificial intelligence and the culture of digital self-help. This article analyzes the risk of patients losing their agency when confronted with algorithms and the asymmetry of clinical power.
The reader will learn how to build a therapeutic relationship based on competent epistemic humility. You will also discover why AI can become a dangerous refuge from genuine contact with another human being.
Power Asymmetry and the Risk of Reducing Humans to Diagnoses
The psychiatrist-patient relationship is exceptionally complex, as it concerns the realm of identity and the interpretation of suffering. Unlike other fields of medicine, here the physician holds the power to define what constitutes a symptom versus a personality trait.
There is a real risk of reducing a human being to a diagnostic label. According to John Steiner's concept of being seen, the critical gaze of the physician can be experienced as a degradation of the self.
Ethical care therefore requires the protection of patient autonomy. The physician must not be the sole arbiter of reality, but should instead strive to see the person behind the symptom.
Self-Diagnosis and Self-Help as Clinical Material, Not a Battleground for Authority
Modern patients often enter the clinic with a ready-made self-diagnosis from the internet or through the use of self-help tools. The psychiatrist should not treat this as an attack on their authority, but rather as valuable clinical material.
An ethical response involves treating such a label as a hypothesis. Rather than simply confirming or rejecting it, the physician should explore what that specific category means to the patient and how it helps them understand their own life.
However, it is important to distinguish between agency and the privatization of health responsibility. Helpful self-help empowers the individual's subjectivity, whereas harmful self-help shifts the entire burden of treatment onto the individual, ignoring the social determinants of health.
AI as an Illusion of Relationship and a Technological Refuge
The use of AI in self-help carries the risk of creating a technological refuge. Generative models simulate empathy, offering a relationship without the costs associated with confronting another distinct human being.
Risks include a lack of real clinical accountability and the danger of reinforcing erroneous beliefs through the mechanism of reassurance seeking. AI can become a golden cage that deepens isolation and dependency instead of reintegrating the patient into the world.
Safety in chatbots is inconsistent; they lack the ability to assess subtle crisis signals. Therefore, AI should be merely a supporting tool, not an autonomous diagnostician or a substitute for the therapeutic bond.
Summary
Psychiatric ethics in the digital age must be based on the ability to see without appropriating. The true success of therapy and technology is measured by whether the patient regains their autonomy and their capacity to be a subject.
The ultimate criterion for the value of any tool remains the question: is it a safe asylum from which one can depart, or a cage that prevents a return to the world? In a world of algorithmic certainty, preserving the right to remain not fully known is essential.
Frequently Asked Questions
Why is the psychiatrist-patient relationship ethically more complex than in other fields of medicine?
This relationship is more complex because the psychiatrist possesses the power of interpretation, which allows them to label a patient's experiences as symptoms or restrict their freedom. Additionally, there is a strong asymmetry resulting from the patient's crisis and dependency, as well as the fact that fear of control or mistrust are often part of the disorder being treated.
1. How should a psychiatrist react to patients who arrive with a self-diagnosis from the internet and use self-help applications?
2. A psychiatrist should not mock a self-diagnosis, but rather treat it as a hypothesis and information about the patient's process of understanding themselves. Their task is to conduct a differential diagnosis and help the patient understand the clinical consequences resulting from various possibilities. In the case of self-help tools, the physician should remember that while they may be effective, they do not replace the therapeutic relationship and clinical responsibility.
3. What are the ethical and psychological risks associated with using AI in psychiatric self-help?
4. The main risks include the risk of intimate data leaks, a lack of real clinical accountability despite the apparent competence of the AI, and the possibility of reinforcing distorted beliefs and avoidance behaviors. There is also a danger of excessive emotional dependence on a tool that may become a substitute for a therapeutic bond and hinder contact with real people.
5. What is the state of safety in generative chatbots designed to support mental health?
6. Generative chatbots supporting mental health employ filters, risk classifiers, model fine-tuning, and design involving clinicians. However, the level of implementation of these safeguards is uneven, which constitutes a fundamental difference from responsible psychiatric care.
7. Where does the helpful role of artificial intelligence in mental health end, and where does the danger to the patient begin?
8. Artificial intelligence is helpful in psychoeducation, organizing symptoms, or practicing emotional regulation; however, it becomes dangerous when it replaces professional diagnosis and clinical decisions. The risk is particularly high regarding pharmacotherapy and in crisis situations, where a reliable risk assessment and real connection to help are essential.
9. Where does helpful self-help and patient agency end, and where does the harmful shifting of responsibility for mental health exclusively onto the individual begin?
10. Helpful self-help increases patient agency, making them a co-participant in treatment through psychoeducation and lifestyle care. It becomes harmful at the moment of the privatization of responsibility, when structural problems (e.g., poverty or violence) are interpreted as errors in the individual's optimization, which can deepen the suffering person's shame.
How to avoid the trap of too quickly and totally defining a patient through a diagnosis or AI algorithms?
One should maintain a multiplicity of hypotheses, accept uncertainty, and recognize that a diagnosis is not identical to a full explanation of a human being. It is important to leave room for complexity and change, and in the case of AI, to be aware that a coherent algorithmic narrative is not necessarily equivalent to the truth.
When is psychiatric and technological assistance ethical, and when does it become a form of domination?
Assistance is ethical when it increases a person's ability to be a subject and allows them to regain agency after its conclusion. It becomes a form of domination the moment it makes the patient or user increasingly dependent on the psychiatrist, application, or AI as the sole arbiters of reality.
What should an ethical doctor-patient relationship look like in a world dominated by internet diagnoses and artificial intelligence?
An ethical relationship should be based on the ability to see the patient without appropriating them, where professional diagnosis meets the patient's self-knowledge without a ritual of domination. The physician should consider the influence of the internet and AI on the patient's perception of symptoms, prioritizing responsibility for the specific individual and the capacity for empathy over the computational speed of systems. It is crucial to respect the patient's right to intimacy and privacy, so that help restores them to reality without defining them entirely by a diagnosis or algorithm.
