1,003 people with depression rated psychotherapy with different levels of AI involvement: they preferred human therapists, willing to pay 31.6% less for assistive or collaborative AI and 57.2% less for fully autonomous AI

People with depression show systematically decreasing acceptance of psychotherapy as AI involvement increases, from a human therapist without AI to assistive, collaborative, and fully autonomous AI. Prior work has largely focused on the technical feasibility or clinical effects of AI in psychotherapy, whereas this study turns to prospective patients' evaluations and distinguishes multiple levels of AI involvement rather than treating AI as a single category. Based on 1,003 participants with depression rating vignettes, with consistent direction across likelihood of seeking treatment, hesitancy, and treatment acceptability, a relatively large sample covering multiple levels of AI involvement.

Resistance to AI-integrated therapy has a quantifiable economic dimension: compared with a human therapist, participants were willing to pay 31.6% less for therapists using assistive or collaborative AI and 57.2% less for fully autonomous AI. Extends acceptance from the attitudinal level to willingness to pay, providing a patient-side quantitative reference for the economic viability of AI psychotherapy. Willingness-to-pay differences are reported as percentages and widen as AI autonomy increases, consistent in direction with the refusal rates rising with autonomy.

A substantial share of participants would outright refuse AI-integrated treatment: 18-19% for assistive or collaborative AI and 31% for fully autonomous AI. Reveals that even at lower levels of AI involvement nearly one in five prospective patients would opt out entirely, while refusal approaches one in three for fully autonomous AI. Refusal rates are self-reported by participants in a sample of 1,003 people with depression, derived from vignette evaluations rather than actual treatment choices.

Preference for human over AI therapists is linked to therapist experience level: it was more pronounced when comparing doctoral or masters-level therapists with peer or trainee therapists. Suggests that patient acceptance of AI is not uniform but interacts with the level of human expertise being replaced. This comparison is based on therapist experience levels set in the vignettes and reflects differences at the level of participant evaluations rather than actual treatment outcomes.

The study addresses a prospective patient group described as having depression and measures their evaluations of vignettes describing psychotherapy with different levels of AI involvement, rather than actual choices or clinical outcomes in real treatment. Its conclusions apply to assessing patient acceptance when mental health services introduce AI, and to service design and pricing discussions, especially when providers consider replacing or assisting therapists of different experience levels with AI. The quantified willingness-to-pay and refusal rates can help providers gauge the market resistance different levels of AI autonomy may face.

Readers may still want to watch: how large the gap is between vignette evaluations and real treatment choices; whether participants' resistance to AI changes with actual experience of use; whether the percentage differences in willingness to pay are stable across income levels or different health payment systems; what mechanism underlies the moderating role of therapist experience; and how the recruitment approach and representativeness of the sample affect generalization. The current text does not include the specific wording of the vignettes, statistical test details, or subgroup analyses, which are open questions to keep in mind when further judging the robustness of the results.