The impact of artificial general intelligence on mental health (a systematic narrative review)

Artificial General Intelligence (AGI)—systems theoretically capable of performing any intellectual task at human level or beyond—is projected by several research communities to emerge before 2030, with profound potential implications for mental health that remain inadequately examined. While AGI promises revolutionary advances in diagnosis and treatment accessibility, it also poses unprecedented risks including mass unemployment, erosion of authentic human relationships, and algorithmic manipulation. This study aims to: (1) analyse the theoretical benefits and risks of AGI on mental health from clinical, ethical, and socioeconomic perspectives; (2) identify regulatory and ethical strategies to mitigate risks associated with behavioural manipulation and inequality; and (3) examine AGI’s anticipated impact on human identity and social dynamics within cultural and occupational contexts. We conducted a systematic narrative review following PRISMA guidelines, operationalised through an explicit PICOS framework (Population, Intervention/Exposure, Comparison, Outcomes, Study design), across five databases (PubMed, Scopus, Web of Science, Dialnet, and PsycINFO) covering publications from 2013 to 2023. Two reviewers independently screened records at the title/abstract and full-text stages, with disagreements resolved by discussion to consensus. Given that AGI does not yet exist as a clinically deployable technology, the eligible literature is predominantly theoretical, ethical, or conceptual in nature; the review therefore adopts an explicit hierarchy-of-evidence framework that differentiates empirical studies, secondary reviews, ethical/normative analyses, and theoretical/speculative work, and weights the synthesis accordingly. Of 72 initially identified records, 45 studies met the inclusion criteria and underwent thematic synthesis using NVivo software, with quality appraised using author-developed criteria explicitly aligned with the SANRA scale for narrative reviews [4] and the AACODS checklist for non-empirical and grey-adjacent sources [35]. Three key thematic findings emerged, each reported with explicit attention to the evidentiary status of the underlying literature. First, AGI-adjacent systems demonstrate potential to support mental health diagnoses through multimodal data processing and to reduce stigma via accessible conversational interfaces (empirically documented for narrow AI), though the literature consistently warns—at the level of ethical and normative analysis—of risks in automated emotional crisis management, diagnostic error, false reassurance, and challenges specific to psychotic and severe mental illness presentations, risks that would be substantially amplified in any future AGI system. Second, emerging regulatory frameworks—including mandatory algorithmic audits, transnational data protection policies, and clearer allocation of medico-legal accountability—are identified as necessary preconditions for safe AGI integration. Third, the intersection of AGI with human identity may generate phenomena theorised in the literature as “existential delaborisation” and the algorithmic reconfiguration of biographical narratives, with attendant risks of perpetuating biases in therapeutic interventions. These constructs are advanced as theoretical frameworks derived from established work-identity and AI ethics literatures rather than empirically documented phenomena. AGI integration in mental health represents a dual theoretical force offering democratised psychological care while posing existential and clinical-safety challenges that require urgent, evidence-differentiated ethical governance. Success depends on regulatory frameworks prioritising transparency, equity, accountability, and privacy, coupled with preservation of authentic human connection as the irreplaceable core of mental health care.