The therapist is being augmented, not replaced
the therapist is being augmented, not replaced: openai’s mentalhealthbench started scoring how models handle crises in september, the american psychiatric association says llms must not be sold as clinicians, and ai check-ins between sessions lifted attendance 5 percent with extra symptom relief on top of therapy. the hour of talk stays human.
the hours between are becoming the product.
Context
OpenAI introduced MentalHealthBench on Sep 23, 2026: an open benchmark of 1,215 synthetic mental health conversations with rubrics written by more than 80 licensed mental health experts from 22 countries.
The American Psychiatric Association's resource document, approved July 2026, says AI must be positioned as augmented intelligence supporting rather than replacing psychiatric expertise. A Nature digital medicine paper (Aug 6, 2026) examined AI-enabled continuous care features and engagement.
The Sep 23 MentalHealthBench date and the expert-built rubrics match OpenAI's post. The APA's 'augmented, not replacing' wording matches its document; 'must not be sold as clinicians' is not seen in the sources read, so unsupported here, not refuted. The 5 percent attendance lift is not seen in the sources read, so unsupported here, not refuted. A separate JMIR study reports greater attendance with an AI therapy support tool.
Related work
- OpenAI: Introducing MentalHealthBench ↗Benchmark launch.
- APA: Artificial Intelligence in Psychiatric Practice ↗Resource document, July 2026.
- npj Digital Medicine: AI-enabled continuous care features ↗Engagement and outcomes.
Watch next
- Look at how MentalHealthBench scores crisis responses specifically.
Sources
Provenance
The note above is reproduced unedited from the original post, first published on Threads on 2 October 2026 at 12:21 IST. Sources are the papers and datasets the note draws on.
View the original post ↗Embed this note
More notes
The air is now being asked to keep its own ledger
the air is now being asked to keep its own ledger: ecmwf’s aifs compo becomes the first ai model to forecast atmospheric composition globally every three hours, cleanair simulates 365 days of pm2.5 over china in ten seconds, and a unified framework maps six pollutants at one kilometer across the whole country. the air now files its own composition report.
read the note →The current is now being asked to draw its own map
the current is now being asked to draw its own map: china’s langya 2.0 predicts six ocean phenomena including internal waves and mesoscale eddies, a deep net called wenhai resolves eddies globally with air sea flux formulas built in, and scripps infers surface currents from the way temperature patterns deform in satellite images. the ocean now files its own circulation report.
read the note →The soil is now being asked to report its own carbon
the soil is now being asked to report its own carbon: a nix color sensor paired with generative data augmentation predicts soil organic carbon without a lab, random forest drives 74 percent of soil health mapping studies, and sentinel 2 tracks five year carbon change across france and italy from 922 samples. the dirt now files its own carbon account.
read the note →