A perspective piece in Solutions Review examines the expanding role of AI in mental health contexts, with particular attention to how the technology is being used to reduce administrative burden for clinicians — including documentation, session notes, and treatment planning workflows.
The author raises a parallel concern: that as AI becomes more conversational and more capable of simulating empathy, the field risks losing sight of what makes therapy effective. The human relationship — the attunement, the non-verbal cues, the genuine mutual presence — is not something that can be replicated by a language model.
The article does not argue against AI in clinical settings, but rather for thoughtfulness about which parts of the work AI should support and which it should not.
This is a useful frame for thinking about appropriate AI use in therapy. The argument isn’t that AI is harmful — it’s that poorly scoped AI use risks substituting for the most irreplaceable parts of the therapeutic relationship.
Tools designed to reduce administrative load — so that clinicians can spend more time on direct care — represent one end of this spectrum. Tools that attempt to replicate the therapeutic relationship itself represent the other.
Understanding where a tool sits on that spectrum matters when clinicians are evaluating what to incorporate into their practice.