AI meets psychology: an exploratory study of large language models’ competence in psychotherapy contexts
Bibliographic record
Abstract
The increasing prevalence of mental health problems coupled with limited access to professional support has prompted exploration of technological solutions. Large Language Models (LLMs) represent a potential tool to address these challenges, yet their capabilities in psychotherapeutic contexts remain unclear. This study examined the competencies of current LLMs in psychotherapy-related tasks including alignment with evidence-informed clinical standards in case formulation, treatment planning, and implementation. Using an exploratory mixed-methods design, we presented three clinical cases (depression, anxiety, stress) and 12 therapy-related prompts to seven LLMs: ChatGPT-4o, ChatGPT-4, Claude 3.5 Sonnet, Claude 3 Opus, Meta Llama 3.1, Google Gemini 1.5 Pro, and Microsoft Co-pilot. Responses were evaluated by five experienced clinical psychologists using quantitative ratings and qualitative feedback. No single model consistently produced high-quality responses across all tasks, though different models showed distinct strengths. Models performed better in structured tasks such as determining session length and discussing goal-setting but struggled with integrative clinical reasoning and treatment implementation. Higher-rated responses demonstrated clinical humility, maintained therapeutic boundaries, and recognised therapy as collaborative. Current LLMs are more promising as supportive tools for clinicians than as therapeutic applications. This paper highlights key areas for development needed to enhance clinical reasoning abilities for effective mental health use.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.001 | 0.000 |
| Research integrity | 0.000 | 0.001 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".