本研究聚焦於發展「人智互動需求架構」，以回應人工智能（AI）技術由輔助工具快速進化為智能夥伴後，現有理論無法全面滿足多元且隨技術演進的人機互動需求之落差。本研究整合 Maslow 需求層次論與 Alderfer 的 ERG 理論，提出人智互動需求架構：將人類在人智協作場景中的需求系統性地解構為「協作需求」、「自然互動需求」、「獨立思考需求」與「人機共融需求」四個層次，開創了人智互動理論研究的人本新視角。並且透過了質性調查確認了此架構存在的合理性。 為驗證此理論架構之實用性與有效性，本研究以生成式 AI（GAI）之技術遷移為實證案例，基於「推-拉-維繫（PPM）」遷移理論開展子研究，透過回收台灣 204 位樣本，運用 SmartPLS 4 進行多階段 PLS-SEM 量化分析。 實證結果顯示，在推力與拉力方面，傳統方式之成本與 GAI 的享樂性顯著提升使用者的認知價值；然而，傳統工具彈性與客製化不足，以及 GAI 的創意與擬人化特質之主效果皆不顯著。此現象反映出高知識群體因成熟的檢索習慣與高度自主控制感，補償了傳統工具的缺陷；同時，由於擔憂 GAI 在執行功利性任務時伴隨的幻覺風險與驗證成本，使用者傾向於在評估時抑制對其發散創意與擬人社交特質的價值認同。在維繫力方面，個人需求滿足與組織情境障礙直接影響認知價值，而個人的科技焦慮與心理不適感則因自然語言介面的極低使用門檻而被技術性稀釋。此外，所有維繫力的交互調節效果全數不顯著，反而支持了維繫力在智慧科技遷移中扮演平行直接路徑而非調節變數的學術論點。 本研究之發現不僅擴展了 PPM 遷移理論與需求理論的學術邊界，更為技術開發廠商之「智能設計」與組織引進 AI 之「變革管理」提供了實務指引。;This research focuses on developing the "Human-AI Interaction Demand Framework" to address the theoretical gap caused by Artificial Intelligence (AI) rapidly evolving from auxiliary tools into intelligent partners. By integrating Maslow′s Hierarchy of Needs and Alderfer′s ERG theory， this study systematically deconstructs human demands in Human-AI Interaction (HAI) into four levels: Collaboration demands， Natural Interaction demands， Independent Thinking demands， and Human-AI Integration demands. This framework presents a human-centered perspective in HAI research， and its foundational rationality was confirmed through a qualitative thematic analysis. To empirically validate the framework′s practical efficacy， a sub-study was conducted within the context of Generative AI (GAI) technology migration based on the Push-Pull-Mooring (PPM) framework. Utilizing 204 valid samples from Taiwan， a multi-stage Partial Least Squares Structural Equation Modeling (PLS-SEM) analysis was executed. The empirical findings indicate that regarding push and pull forces， the costs of traditional methods and the hedonism of GAI significantly enhance users′ perceived cognitive value. Conversely， the main effects of traditional tools′ lack of flexibility/ customization and GAI′s creativity and anthropomorphism are non-significant. This phenomenon reflects that high-knowledge demographics compensate for traditional tool defects through mature information retrieval habits and a strong sense of autonomy. Furthermore， due to concerns over hallucination risks and verification costs in utilitarian tasks， users tend to suppress value recognition of GAI′s creative and anthropomorphic traits. Regarding mooring forces， personal need satisfaction (dominated by existence needs) and contextual barriers directly impact perceived cognitive value. Meanwhile， personal technology anxiety and discomfort are technically diluted by the extremely low entry barrier of natural language interfaces. Crucially， all moderating interaction effects of mooring forces are non-significant， supporting the academic argument that mooring acts as parallel direct paths rather than moderating variables in smart technology migration. The findings extend the theoretical boundaries of PPM migration theory and need theories while offering practical guidelines for technology developers in "intelligent design" and organizations in AI-driven "change management".
