AI-mediated relational competence and its limits: Psychedelic-assisted therapy as a stress case and policy signal
Robert McGrath, Everett B. Sackett
Psychedelics August 14, 2026 DOI: 10.1016/j.psyche.2026.100035 (opens in new tab)
Study at a glance
AI-extracted from the abstract| Characteristics | Theoretical or philosophical paper Peer reviewed |
|---|---|
| Topics | Psychedelic-assisted therapy |
| Key points | Argues that the AI-mediated relational competence construct's empathic-presence and relational-judgment components hold and intensify in psychedelic-assisted therapy, while trustworthiness, transparency, accountable-decision-making, and equity-aware components strain against the altered-consciousness context. Raises a policy question about private AI platforms becoming de facto certification standards. |
Abstract
Artificial intelligence is entering psychedelic-assisted therapy (PaT) through at least two channels, pre-licensure facilitator training simulators and in-session administrative tools. This arrival is occurring in a context where relational factors, those interpersonal considerations between patients and providers, are not incidental but highly impactful. A 2025 international Delphi consensus study formally recognized therapeutic alliance, participant trust, and study personnel qualifications as reportable determinants of psychedelic trial outcomes. This paper uses PaT as a limit case for AI-mediated relational competence, a six-part construct originally developed for the standard clinical encounter, to test whether it holds, strains, or requires revision under conditions of altered consciousness and heightened vulnerability. Two live examples anchor the analysis: an AI simulation platform for facilitator training and ambient-AI documentation tools now used in ketamine-assisted psychotherapy. The construct's empathic-presence and relational-judgment components largely hold, and are arguably intensified, in this setting. Trustworthiness, transparency, accountable-decision-making, and equity-aware components strain against PaT's altered-consciousness context, as does its fragmented accountability infrastructure and its own documented history of racial and cultural inequity. The paper closes with a policy question. Absent an accreditation infrastructure comparable to accredited medical education, should a privately developed AI platform become the de facto standard for facilitator competency certification?