Why AI use decays after adoption in hearing care: stabilising AI-enabled knowledge in clinical practice

Purpose This study aims to examine why artificial intelligence (AI)-enabled innovations may experience decay in use after initial implementation. Specifically, it aims to identify stage-specific barriers across abandonment, scale-up, spread, and sustainability, with particular attention to why AI-enabled knowledge may fail to stabilise in routine clinical practice. Design/methodology/approach The study adopted a qualitative design, drawing on clinicians’ lived experiences with AI-enabled innovations in hearing care. Data collected through the open-ended essay instrument were analysed using an inductive-abductive logic, with respondent-level codes developed from the data and interpreted through the non-adoption, abandonment, scale-up, spread, and sustainability (NASSS) framework. Findings The study identified NASSS-informed barriers that shape the post-adoption trajectory of AI-enabled innovations in hearing care. At the user/clinical level, abandonment is driven by user-technology interaction barriers, including technical friction, digital burden, lack of trust, and perceived risks. At the organisational level, scale-up is impeded by infrastructure and system integration constraints, as well as capacity and change-management load. Next, spread is hindered by disparities in resources and readiness across settings, as well as the dependence on supporting conditions to enhance wider transferability and legitimacy. Finally, sustainability is challenged by ongoing support requirements and usage burdens, and user hesitancy and preferences. Originality/value The study contributes by reframing post-adoption decay in AI use as a knowledge breakdown problem, where AI-enabled knowledge may remain fragile after initial uptake. It interprets abandonment as a confidence breakdown, scale-up barriers as embedding breakdowns, spread barriers as transfer breakdowns, and sustainability barriers as engagement breakdowns. It also develops a post-adoption barrier matrix that translates these breakdowns into a practice-oriented structure, helping clinicians and healthcare organisations understand why AI use decays, where the barriers are located, when they become salient, why AI-enabled knowledge breaks down, and the risks this poses to continued use.