A Guide for Trustworthy AI Implementations in the Smart Living Domain

Abstract

Advances in AI facilitate the development of numerous smart solutions that significantly enhance efficiency and comfort for organizations and individual users. However, AI systems often suffer from the 'black box' problem and other issues that challenge their trustworthiness, and thus broad user acceptance. Since trust is crucial for the initial adoption and continued engagement with AI systems, there are numerous academic and practice-driven efforts to articulate requirements for trustworthy AI applications. The sheer volume of frameworks and information can be overwhelming. This article distills prior literature and practice insights and presents a structured guide outlining the main requirements for trustworthy AI applications. While we focus on the smart living domain, our aim is to provide concrete guidance for development and reduce the complexity of the topic for SMEs seeking to develop AI applications in various application fields.