Psychologist

A major test publisher's announcement. It says the assistant draws assessment data from its Q platforms, combines it with clinician-provided notes and documents, and generates draft narrative sections of psychological and educational evaluation reports; that clinicians remain fully responsible for reviewing and editing content, interpreting assessment results and finalising reports; and that it aims to reduce report-writing time by 50% or more. It is the company describing its own product and its aims, not measured outcomes.

A demo, benchmark or paper shows the task can be done. Updates what the technology can do — not what employers will do.

Australia. A mixed-methods study with 12 interviews and a national survey of 278 mental health professionals recruited by convenience sampling; 84 respondents (30.2%) were psychologists, the rest counsellors, social workers, occupational therapists, psychiatrists and others. 121 of 278 (43.5%) reported daily administrative or clinician-support AI use and 92 of 278 (33.1%) daily client-facing AI use, with concerns about privacy, accuracy and uneven workplace governance. A convenience sample of several professions.

Measured, large-scale use of a tool for real work, where the decision to use it was the worker's rather than an employer's. It is more than a capability record — the work is real, not a demo — and less than a deployment record, because no employer put it into production, required it, or built a process around it. Weighted `cautious`: `automating` means the machine can do the task AND there are adoption signs, and this is an adoption sign — but usage can be experimental, and much of the measurement comes from a party with a stake, so one record is never enough and two independent ones are. Note who is counting. Vendor telemetry sees this directly and sells the tool, so such a record names that stake in its scope; a statistics agency asking firms whether their workers use AI in tasks sees the same channel with no stake at all, and that is the better source where it exists.

United States. The statistics bureau projects overall psychologist employment to grow 6 percent from 2025 to 2035, with about 11,300 openings a year; clinical and counseling psychologists from 81,300 to 90,900 (12 percent) and school psychologists from 64,000 to 64,500 (1 percent). It attributes growth to demand for psychological services and says declining school enrollment may limit demand for school psychologists; the page does not mention artificial intelligence. It counts jobs for one country.

A named person with standing publicly predicted something, on a date, in an attributable statement. It is recorded so that who said what, and when, stays checkable — and it never moves a task's assessment, because a prediction is not an observation. Its value arrives later: the record sits on the same page as the evidence about that occupation, so anyone reading the forecast reads the record of what happened next beside it. That is the reckoning; this site publishes no verdict on whether a forecast came true.

Australia's professional body for psychologists, summarising its AI and emerging technologies practice guidelines. It says psychologists remain solely responsible for all outcomes and AI is an assistant, not a replacement for professional judgement; generic AI disclosures are insufficient and the guidelines require explicit consent for each specific tool and data use; AI outputs are part of the clinical record. A professional body's guidance, not a regulator's rule; the full guidelines are for members and were not read.

Failure, rollback, regulation or cost is suppressing adoption. Can lower an assessment or widen its uncertainty.

United States. APA's annual practitioner survey, completed by 1,742 psychologists in September 2025 from a probability-based random sample, found 56% had used AI tools to assist with their work at least once in the past 12 months (29% in 2024) and 29% at least monthly. Among users, the most common uses were writing emails and other materials (52%), generating content (33%), summarising clinical notes or articles (32%) and note-taking (22%); 8% used AI to assist with clinical diagnosis and 5% chatbot assistance for patients. 38% worried AI may make some of their duties obsolete. Self-reported use; it does not measure outcomes.

Measured, large-scale use of a tool for real work, where the decision to use it was the worker's rather than an employer's. It is more than a capability record — the work is real, not a demo — and less than a deployment record, because no employer put it into production, required it, or built a process around it. Weighted `cautious`: `automating` means the machine can do the task AND there are adoption signs, and this is an adoption sign — but usage can be experimental, and much of the measurement comes from a party with a stake, so one record is never enough and two independent ones are. Note who is counting. Vendor telemetry sees this directly and sells the tool, so such a record names that stake in its scope; a statistics agency asking firms whether their workers use AI in tasks sees the same channel with no stake at all, and that is the better source where it exists.

One US state, announced by the agency that licenses psychologists and other behavioural health professionals. It says the Wellness and Oversight for Psychological Resources Act prohibits anyone from using AI to provide mental health and therapeutic decision-making, while allowing AI for administrative and supplementary support services for licensed behavioral health professionals. It does not bind other jurisdictions, and an announcement is not the statute's text.

Failure, rollback, regulation or cost is suppressing adoption. Can lower an assessment or widen its uncertainty.

England. The health technology body says two digital front door technologies, Limbic Access and Wysa Digital Referral Assistant, can be used in the NHS during an evidence-generation period to gather service user information for NHS Talking Therapies assessments, only while the evidence is being generated and once they have regulatory approval; after three years NICE will assess whether they can be routinely adopted. It governs information gathering before assessment, not the assessment itself.

Failure, rollback, regulation or cost is suppressing adoption. Can lower an assessment or widen its uncertainty.

A study in which 249 licensed psychologists evaluated psychological reports written by human psychologists and by ChatGPT-4 on quality, readability, style, organisation, summaries, recommendations, preference and willingness to sign off. Human reports were generally rated more favourably and raters were more comfortable approving them, but effect sizes were typically small; human summaries were moderately better, and AI reports moderately better on recommendations. It rates reports, not practice.

A demo, benchmark or paper shows the task can be done. Updates what the technology can do — not what employers will do.

Japan, the statute for the national psychologist qualification, in the version in force from 1 June 2025. Article 2 defines the certified psychologist's work as observing and analysing the psychological state of people needing support, counselling and advising them and those around them, and education. Article 42(2) requires the certified psychologist, where the person has an attending physician for that support, to take that physician's instruction. Article 44 bars non-certified people from using the title or the characters 心理師 — it restricts the title, not the activities. The Act names no technology and counts nobody.

Failure, rollback, regulation or cost is suppressing adoption. Can lower an assessment or widen its uncertainty.

A lab study that mapped therapy guides used by major medical institutions and tested current language models such as gpt-4o against them. It reports that the models express stigma toward those with mental health conditions and respond inappropriately to certain common and critical conditions, including encouraging clients' delusional thinking, even with larger and newer models. It tests models, not deployed products or clinicians.

A demo, benchmark or paper shows the task can be done. Updates what the technology can do — not what employers will do.

A randomised trial of 106 US adults diagnosed with major depressive disorder, generalized anxiety disorder or an eating disorder, reported by the investigators' own institution. Participants with depression reported a 51% average reduction in symptoms and those with anxiety 31%; the investigators say no generative AI agent is ready to operate fully autonomously in mental health. The comparison was with people who did not use the chatbot, not with therapists, and the software was developed by the same lab.

A demo, benchmark or paper shows the task can be done. Updates what the technology can do — not what employers will do.

England. A multisite observational study of 129,400 patients in NHS services found that services using an AI-enabled self-referral chatbot saw referrals rise 15%, against 6% in control services, with larger increases among minority groups. The chatbot gathers information at referral, before a clinician's assessment. Five authors are employed by and hold shares in the company that makes it, and another is a paid consultant; it is peer-reviewed but not independent.

An employer has put it into production. Can move the baseline — weighted by scale and how similar the setting is.

If you are training, expect AI in your notes and report drafts from the start, and build what it cannot do for you: administering tests properly, interpreting results against a person's history, and conducting therapy under supervision.

Expect report drafting and paperwork to speed up, and to answer for every line you sign. Diagnosis and therapy stay with you, and in some places the law now says so.

Reshape the role

Use AI for paperwork, keep interpretation and diagnosis

Surveys show most use is administrative, and professional guidance keeps psychologists solely responsible for outcomes.

Stay and strengthen

Stay in assessment and therapy where a licence is required

Rules in at least one state bar AI from therapeutic decisions, and a trial's own investigators say AI is not ready to operate autonomously.

Adjacent move

Move into evaluating AI tools for mental health services

England's health technology guidance requires evidence to be generated on AI intake tools, and that work needs clinical judgement.

The evidence does not show that. Most psychologists' AI use is paperwork, report drafting is being assisted, and diagnosis and therapy remain with the psychologist — at least one US state bars AI from therapeutic decisions. The US projection expects psychologist employment to grow 6 percent from 2025 to 2035, without mentioning AI.

We do not answer that with a number of years. Watch whether your services buy AI intake or report tools, and what your licensing rules say about AI in therapy. Those tell you more than any date.

A trial of one therapy chatbot reported symptom reductions against no treatment, but its own investigators say no generative AI is ready to operate autonomously, a study found models encouraging delusions, and Illinois bars AI from therapeutic decision-making.

It can draft them. Licensed psychologists rated GPT-4 reports only slightly below human ones, and a test publisher now drafts report sections in its platforms — but the clinician remains responsible for interpreting the results and signing the report.