Therapeutic AI Francophone

Gaëlle Charlot

Registered Occupational Therapist, M2 Applied Philosophy in Psychiatry

At a glance: Gaëlle Charlot is an occupational therapist at the SMPR in Bordeaux-Gradignan (Centre Hospitalier Charles Perrens), specializing in forensic psychiatry for over 20 years. A pioneer in the therapeutic use of generative AI in correctional settings, she uses Suno AI (music creation) and ChatGPT (conversational support) as therapeutic mediations with incarcerated patients. Her concept of the “non-perceptible third” opens up an unprecedented reflection on the role of AI in the therapeutic relationship.

Profile

Institution: Service Médico-Psychologique Régional (SMPR), Centre Pénitentiaire de Bordeaux-Gradignan — Centre Hospitalier Charles Perrens

Education: Registered Occupational Therapist (2002); M2 “Care, Ethics and Health”, Université Bordeaux Montaigne (supervisor: Steeves Demazeux, 2023). HAS (French National Authority for Health) expert.

Career: 24 years of experience in psychiatric occupational therapy, spanning four institutional contexts — rural psychodynamic setting (Picardy), CBT-oriented approach (Pontoise), university hospital (Sainte-Anne, Paris), and for several years the correctional setting in Bordeaux. This diversity of practice settings nurtures a rare integrative approach.

Research Area

Gaëlle Charlot's work sits at the intersection of occupational therapy, philosophy of psychiatry, and generative AI. Three main themes emerge.

Generative AI as therapeutic mediation

Charlot integrates Suno AI (music creation from keywords) into her occupational therapy sessions with incarcerated patients. The advantage: an “immediate output” that avoids the failure experience of traditional mediations (painting, clay modeling) with patients who often have cognitive deficits. Varying musical styles for the same lyrics allows work on perception, frustration tolerance, and emotional differentiation.

The “non-perceptible third”

An emerging concept: AI constitutes a new type of therapeutic mediator, neither a material object (clay, wicker) nor a human subject (therapist), but a generative agent that co-produces therapeutic material in unpredictable ways. This unpredictability is therapeutically productive: it creates tolerable frustration and requires the patient to position themselves in response to surprise.

Musical catharsis and emotional regulation

Her M2 thesis (supervisor: Steeves Demazeux) philosophically grounds her practice: Aristotelian catharsis applied to emotional regulation through music with violent patients. Listening to music congruent with one's emotional state (metal for anger, sad music for grief) produces regulation more accessible than traditional mediations.

Why this matters for clinicians

Gaëlle Charlot is not a researcher theorizing about AI from a lab. She is a field practitioner who uses generative AI in sessions, with some of the most vulnerable patients (incarcerated, substance users, psychotic). This profile is rare and valuable for three reasons.

Concrete protocols: she has developed pre/post-session questionnaires to document the effects of Suno AI on emotional expression. This is documented clinical work, not anecdote.
Philosophical reflexivity: her training with Steeves Demazeux (philosophy of psychiatry) gives her a capacity for critical analysis that most AI-enthusiast practitioners lack.
Extreme populations: correctional settings pose radical constraints (limited access, vulnerability, risk of acting out). If therapeutic AI works there, the lessons are transferable to less constrained contexts.

Key Publications

Occupational Therapy Towards a Modern and Digital Approach to Care — Suno AI (2025)

Foundational article describing the protocol for integrating Suno AI into correctional occupational therapy. Includes methodology, questionnaires, and preliminary results.

ResearchGate

Anger: Catharsis or Pathology? (M2 Philosophy, 2023)

M2 thesis (supervisor: Steeves Demazeux, Bordeaux Montaigne). Theoretical foundation for emotional regulation through music: articulation between Aristotelian catharsis and clinical practice with violent patients.

The Role of Healthcare Professionals Facing AI-Induced Psychosis (2025)

Reflection on psychotic risks related to AI in vulnerable populations. Complements her approach by integrating vigilance dimensions.

Webinar “Music, AI and Correctional Settings” (GRESM, 2025)

First webinar dedicated to AI-generated music in correctional settings. Presentation of work with Suno AI and exchanges with healthcare professionals.

Related Concepts on This Site

Gaëlle Charlot's work directly intersects several concepts documented in our resources:

Computational Creativity

Margaret Boden's framework directly illuminates what Charlot's patients do with Suno AI: the P-creativity / H-creativity distinction shows that novelty for the patient is the only clinically relevant criterion.

Therapeutic Computational Creativity

Therapeutic extension of Boden's framework, directly inspired by Charlot's practices: how computational creativity becomes a clinical tool.

Complementary Theoretical Framework

Charlot's work gains depth when read in dialogue with other frameworks.

Margaret Boden — Computational Creativity

Boden's framework (1990) provides the theoretical vocabulary to analyze what Charlot observes empirically: the three types of creativity (combinational, exploratory, transformational) and the P/H-creativity distinction structure the understanding of AI-assisted creative processes.

Winnicott — Transitional Space

Winnicott's concept of the transitional object (neither me nor not-me) offers a psychodynamic framework for thinking about Charlot's “non-perceptible third”: AI creation as a new transitional space between inner world and reality.