LO
CapabilityCognitive automation2025-03-27
A generative AI therapy chatbot that participants could message whenever they needed to talk cut depression symptoms 51% in a trial, but no AI is ready to run unsupervised, its developers said
Counsellor / therapistoccupation page →
Event date / reported
2025-03-27
Evidence stage
CapabilityA demo, benchmark or paper shows the task can be done. Updates what the technology can do — not what employers will do.
Tasks this bears on
Being there at three in the morning
Low-intensity support at the moment it is wanted: someone is awake, anxious, and there is no appointment for eleven days.
Automating✓ Evidence-backed
Making a therapeutic decision
Deciding this person is at risk, that the plan has to change, or that they need someone other than you.
Still human-led✓ Evidence-backed
Where this applies
A randomised trial of 106 US adults with major depressive disorder, generalized anxiety disorder or an eating disorder, who used a smartphone chatbot by responding to prompts or initiating conversations when they needed to talk. Depression symptoms fell 51% on average and anxiety 31%, and participants reported trusting the system to a degree comparable to working with a mental health professional; the investigators say no generative AI agent is ready to operate fully autonomously in mental health. The comparison was with people without the chatbot, not with counsellors, and the lab built the software.
What this means
The availability counsellors cannot offer is where a chatbot showed benefit in a trial — with its own makers insisting it still needs supervision.
What it does not yet show
One trial against no treatment, run by the software's developers; it does not compare the chatbot with counselling.
What you can check
Open Dartmouth's release "First therapy chatbot trial yields mental health benefits" and find "when they needed to talk".
Does it change the assessment?
No — and this stage does not move it either. A "Capability" record is real evidence, but it does not upgrade a task judgement on its own. The 2 linked judgements above stand where they were.
Source
Dartmouth College — news release on the Therabot trial published in NEJM AI (3/27/2025) · verified 2026-09-30 · Claude (VOLO agent) · interpreted 2026-09-30 · Claude (VOLO agent)
Primary source — published by the party that did this, or the authority of record. No co-signature needed.