Responsible artificial intelligence in long-term care: From concept to context

Abstract

Worldwide, advancements in artificial intelligence (AI) are widely positioned and predicted to spur societal and economic progress and help address today’s pressing challenges such as the growing demands of an ageing population. In long-term care (LTC) for older adults, AI enables technologies such as remote monitoring systems, decision support systems, socially assistive robots, and virtual assistants. Such technologies can enhance care quality, while also easing pressure of the growing caregiver shortage by streamlining processes and improving efficiency. However, alongside the opportunities of AI technologies come societal and ethical concerns, such as the erosion of privacy, autonomy and interpersonal interactions, and the exacerbation of bias and opacity. It is therefore essential that those developing and using AI technologies in LTC endorse responsibility. This includes carefully balancing the promises and benefits of AI technologies against their risks and disadvantages, engaging deeply with societal and ethical implications, and shaping AI technology development and deployment in alignment with the values and needs of older adults, their formal and informal caregivers, and society as a whole. This dissertation explores how responsible AI innovation is practically enacted in LTC for older adults. Two conceptual perspectives provide the foundation for this dissertation: the discourse on responsible AI, which stresses high-level principles such as transparency, accountability, justice, and human autonomy; and the field of Responsible Innovation (RI), which emphasizes the process-oriented AIRR principles — anticipation, inclusion, reflexivity, and responsiveness. Both perspectives are integrated to allow for a dynamic approach to exploring the process- and outcome-oriented aspects of responsible AI innovation in LTC for older adults. Drawing from empirical and conceptual findings, this dissertation identifies three overarching and interconnected pathways for responsible AI innovation in LTC for older adults, namely: 1) proactive balancing of AI’s opportunities and risks; 2) fostering user-centric learning to guide responsibility by design, configuration, and implementation; and 3) reconciling context-sensitivity with scalability in AI innovations for diverse care settings. These pathways can be regarded as a broad synthesis of the findings that bridges theory, empirical evidence, and practical recommendations. They include strategies, frameworks, and decision-making processes that can facilitate the responsible development and deployment of AI technologies in alignment with the specific values and needs of users and other stakeholders. While not exhaustive, the three identified pathways capture core factors that are likely to shape whether AI in LTC contributes to outcomes that are ethically acceptable and socially desirable. This thesis lays a foundation for critically engaging with responsible innovation in AI and LTC, offering both theoretical depth and practical guidance, with relevance that may extend to other technological and social innovations and other sectors.