The benefits and future potential of generative artificial intelligence (GAI) on mental health: a Delphi study.

Statistically significant consensus was achieved among experts regarding both the benefits and future potential of generative AI in mental health care.

• Consensus was measured using Kendall's coefficient of concordance (W).
• Benefits consensus: W = 0.145, p = 0.034.
• Future potential consensus: W = 0.152, p = 0.025.
• Both results were statistically significant at the p < 0.05 level.
• A two-round Delphi study was conducted with 15 purposively selected experts.

Twenty-eight themes were identified across eight benefit dimensions of generative AI in mental health.

• Themes were identified through thematic analysis of open-ended responses in Round 1.
• The eight benefit dimensions were ranked by experts in Round 2.
• Accessibility and availability ranked as the most important current benefit dimension.
• Experts were drawn from psychiatry, clinical psychology, counselling, and digital mental health fields.

Twenty-nine themes were identified across eight future-potential dimensions of generative AI in mental health.

• Themes were derived from thematic analysis of Round 1 open-ended expert responses.
• AI as a collaborative and informative tool was ranked as the most important future-potential dimension.
• Experts ranked these dimensions in Round 2 of the Delphi process.
• Future potential dimensions were assessed with statistically significant consensus (W = 0.152, p = 0.025).

Experts perceived generative AI as a transformative adjunct to mental health practice, particularly in expanding access, supporting personalised care, and augmenting professional capacity.

• GAI was not viewed as a replacement for clinicians but as an adjunct tool.
• Key perceived benefits included expanding access to mental health services.
• Personalised care and augmentation of professional capacity were also highlighted as key benefit areas.
• The Technology Acceptance Model (TAM) was used to interpret expert perceptions.

Adoption of generative AI in mental health was identified as contingent on usability, transparency, trust, and robust ethical governance.

• Experts identified these factors as necessary conditions for equitable and human-centred integration.
• Ethical governance was specifically highlighted as a prerequisite for responsible adoption.
• Transparency and trust were identified as key adoption-related concerns alongside usability.
• These findings were interpreted through the lens of the Technology Acceptance Model (TAM).

The study employed a two-round Delphi mixed-methods design with 15 purposively selected experts.

• Experts were selected from psychiatry, clinical psychology, counselling, and digital mental health.
• Round 1 used open-ended questions with responses subjected to thematic analysis.
• Round 2 involved expert ranking of identified themes and dimensions.
• Purposive sampling was used to ensure relevant domain expertise across participants.