Psychologist — tasks, one by one

Giving standardised tests

Administration means a person present under a test's standard conditions, watching how the answers are reached as well as what they are. The US statistics bureau describes clinical and counselling psychologists as interviewing clients, administering diagnostic tests and providing psychotherapy, and projects their employment to grow 12 percent from 2025 to 2035; it does not mention AI.

No record on this page measures AI administering or scoring psychological tests; the judgement is about the structure of the work, and the projection counts jobs in one country.

Writing the assessment report

This is where AI has reached the psychologist's own desk. In a study where 249 licensed psychologists rated reports, human-written ones were generally preferred but the differences were typically small, and AI reports scored better on recommendations. A major test publisher launched a tool in its scoring platforms that drafts narrative sections of evaluation reports, saying clinicians remain fully responsible for interpreting results and finalising reports.

One rating study of GPT-4 reports and a vendor's description of its own product; neither measures report quality in practice or psychologists' workloads.

Intake and diagnosis

Chatbots are already collecting information before a clinician's assessment. England's health technology body says two such tools can be used in NHS Talking Therapies during an evidence-generation period, and a study of 129,400 patients found services using one saw referrals rise 15 percent against 6 percent in control services. Diagnosis itself stays with the psychologist: in a 2025 US survey, 8% of psychologists who used AI said they used it to assist with clinical diagnosis.

The intake evidence is England's talking-therapy services rather than psychologists specifically, and the deployment study was written by the vendor's employees; the survey records use, not accuracy.

Psychotherapy

The rules and the evidence point the same way. Illinois bars anyone from using AI to provide mental health and therapeutic decision-making while allowing administrative support. In the first trial of a generative AI therapy chatbot, participants with depression reported a 51% average reduction in symptoms, yet its investigators say no generative AI agent is ready to operate fully autonomously. A study found language models express stigma toward people with mental health conditions and respond inappropriately, including encouraging delusional thinking. Only 5% of psychologists who used AI in 2025 said they used chatbot assistance for patients.

One state's law, one trial against no treatment rather than against therapists, and one lab study; none measures what happens to psychologists' caseloads.

Notes, letters and paperwork

This is where most use already is. In APA's 2025 survey, 56% of psychologists had used AI tools at least once in the past year, up from 29% in 2024, and the most common uses were writing emails and other materials, generating content, summarising notes or articles and note-taking. An Australian survey of mental health professionals found 43.5% used AI daily for administrative or clinician-support tasks. The Australian Psychological Society says psychologists remain solely responsible for all outcomes and need explicit consent for each specific tool.

Self-reported use in two surveys, one of them a convenience sample of several professions; neither measures time saved or errors in AI-drafted notes.