Design and Evaluation of High-Quality Symbiotic AI Systems through a Human-Centered Approach

Artificial Intelligence (AI) is significantly impacting foreseeing fields offering a better-automated decision making process and autonomous systems. Therefore, it is important to design high-quality AI systems that focus on the users’ priorities and avoid potential unethical and undesired behaviours. In the current scenario, Human-Computer Interaction (HCI) and AI are not separate fields but they contaminate each other and, consequently, the symbiosis between humans and AI system is fostered. Ensuring that AI development benefits humans, while providing an high-level automation, remains a primary concern. For this reason, the human-centered design approach should be adopted to create systems that being trustworthy, safe, reliable, and governance compliant are able to enhance the user’s cognitive abilities and protect they from potential risks. In this context, it is fundamental to identify guidelines to follow while designing high-quality Symbiotic AI (SAI) systems and metrics for their appropriate evaluation. Assessing the empirical validity of the proposed solution is of crucial importance and the planning and execution of a user study is one of the main aspects of this work. The research project concerns the design of SAI systems, more specifically the definition of best practices and metrics to adopt while creating and evaluating these systems. This contribution presents the preliminary results obtained during the initial part of the research. The main opportunities and challenges in this new research field are also discussed.