Vocabulary acquisition is crucial for language learning, yet learners face substantial challenges in memorizing extensive vocabulary. Numerous studies suggest Artificial Intelligence (AI)-based technologies could significantly improve vocabulary acquisition among K-12 learners. Therefore, instructors or teachers must be fully informed of the advantages and disadvantages when integrating different AI apps or programs. The present study provided a systematic review of 30 empirical studies focusing on the use of AI for vocabulary acquisition. Some key findings include: (1) AI-supported L2 vocabulary acquisition (AISVA) research has gained rapid growth over the last decade, with a clear upward trend; (2) Despite AISVA's broad interest across 28 journals, the field shows fragmentation and lacks a unified theoretical framework; (3) The top three most cited in AISVA predominantly employed quantitative methods, concentrate on cognitive learning dimensions with rather innovative theoretical frameworks, and unique research topics; (4) Research methods primarily integrated both qualitative and quantitative approaches, leaning towards a preference for quantitative methodologies, and mainly involved primary and college students; (5) A substantial portion of AISVA applications manifests their roles as "Intelligent Tutors" while the category of "Data-driven Optimizer" has received relatively less attention in research endeavors; (6) Studies emphasize cognitive and affective aspects, with AR and VR emerging as prevalent technologies that are frequently enhanced by AI integration to support vocabulary acquisition. Suggestions and implications for teaching practitioners, language learners, and software engineers were discussed.