Abstract: The paper employ literature review to address theoretical questions related to user innovation under the paradigm of artificial intelligence (AI). It begins by defining AI and user innovation and then conducts a literature search, using specific keywords and inclusion criteria to identify relevant studies. The search strategy is designed to be comprehensive yet accurate, focusing on peer-reviewed studies published in English from 1991 to 2023. The authors use the Gioia methodology for data analysis. This methodology involves three stages: first-order categories, second-order themes, and overall dimensions, moving from a strict focus on empirical data to developing theoretical insights. The purpose of this review is to bridge the gap between AI and user innovation research by identifying key themes and constructing an integrated framework that delineates the reciprocal impact of AI on user innovation and vice versa. The review aims to answer research questions such as "What impact does AI have on user innovation? Why and how can users improve or even innovate AI?" and to pave the way for future research focusing on the interplay of AI, enterprises, users, and multi-actor engagement. The study identifies eight key themes and constructs a comprehensive research framework that provides a systematic understanding of AI and user innovation. The study found that the paths through which AI acts on user innovation include: ①Enterprises identify user innovations through AI; ② AI-driven innovation toolkits; ③AI as a medium reshaping enterprise-user co-creation relationships; ④AI empowering innovation communities; ⑤AI directly participating in innovation as an actor. The paths through which user innovation acts on AI include: ①Consumer users through participatory design. ②Optimizing AI systems through the "human-in-the-loop" framework; ③Enterprise users driving AI innovation through data assets, scenario applications, and multiple roles. The paper enriches existing research results and promotes further research on the bidirectional impact of artificial intelligence and user innovation. It also identifies gaps in the literature and proposes a research agenda to advance innovation in AI users, contributing to the broader literature on user innovation management and providing a foundation for managing AI effectively in enterprises and user innovation processes. We synthesise these findings into an integrative framework that depicts AI and user innovation as a mutually constitutive system. Building on this synthesis, we propose a future research agenda that addresses five under-explored areas: At the firm level, future inquiry should shift from simply using machine learning to identify user insights toward investigating how retrieval-augmented generation and other generative-AI techniques can actively search for, synthesise and integrate user-generated knowledge and solutions. This includes clarifying the new boundaries and governance mechanisms that distinguish AI-native innovation communities from traditional digital ones, and critically examining data-acquisition M&A strategies—such as large firms buying hospitals to secure private health data—by unpacking their strategic motives and the attendant ethical risks of commodifying sensitive patient information. At the user level, in-depth case studies and experiments are needed to tease out the contextual contingencies of AI-driven innovation toolkits, moving AI from a passive “resource” to a co-evolving network actor. Particular attention should be paid to micro-level mechanisms through which (1) employees undergoing digital transformation tinker with or augment AI systems in situ, and (2) individual contributors in open AI communities iteratively refine algorithms via pull requests, issue tracking and human-in-the-loop feedback loops. At the multi-actor ecosystem level, research must go beyond marketing-oriented value co-creation to explore how AI infrastructures mobilise cross-domain collective intelligence in real time. Empirical studies should examine how heterogeneous actors dynamically orchestrate AI capabilities to sense, seize and scale transient digital opportunities within complex, rapidly changing environments.