Research on artificial intelligence and machine learning in mental health has expanded rapidly, yet a field-level synthesis of publication patterns and themes remains limited. We conducted a bibliometric analysis of Scopus records published between 2014 and November 2024. The search retrieved 8,532 records. To enhance relevance and comparability, we restricted to English-language articles and reviews, yielding a final corpus of 1,019 documents. Analyses were performed with the Bibliometrix Shiny application in R, including descriptive indicators (annual output, source impact, frequent terms) and science-mapping (keyword co-occurrence and thematic mapping). Publications increased steadily across 2014–2024 with a marked recent acceleration. The literature concentrates on diagnosis/prediction tasks, with recurring themes in digital phenotyping, language-centric analytics, and conversational agents. Post-2022 records show a visible shift toward generative artificial intelligence. The outlet profile is interdisciplinary, spanning psychiatry, digital-health, generalist open-access, and engineering journal. Artificial intelligence in mental health is at an inflection point, moving from feasibility toward clinical translation. Priorities include adequately powered prospective studies, transparent and standardized reporting, embedded safety/auditability for high-risk contexts, and broader inclusion beyond English-language scholarship. Limitations include reliance on a single index and language/document-type filters.