AI-Assisted Ethics? Considerations of AI Simulation for the Ethical Assessment and Design of Assistive Technologies

• DOI: 10.48550/arxiv.2305.00566
• Publication Date: 2023-01-01
• Authors: Silke Schicktanz; Johannes Welsch; Mark Schweda; Andreas Hein; Jochem W. Rieger; Thomas Kirste
• Keywords: person with dementia; FOS: Computer and information sciences; ethical reflection; intelligent assistive technology; Genetics; Computer Science - Multiagent Systems; QH426-470; simulation; conceptual approach; 3. Good health; Multiagent Systems (cs.MA)

Current ethical debates on the use of artificial intelligence (AI) in health care treat AI as a product technology three ways: First, by assessing risks and potential benefits currently developed AI-enabled products with checklists; second, proposing ex ante lists values seen relevant for design development assisting technology, third, promoting to moral reasoning part automation process. Subsequently, we propose fourth approach AI, namely methodological tool assist reflection. We provide concept an AI-simulation informed separate elements: 1) stochastic human behavior models based behavioral data simulating realistic settings, 2) qualitative empirical value statements regarding internal policy, 3) visualization components that aid understanding impact changes these variables. The this is inform interdisciplinary field about anticipated challenges or trade-offs concrete settings and, hence, spark re-evaluation implementation plans. This may be particularly useful applications deal extremely complex limitations communication resources affected persons (e.g., dementia cognitive impairment). Simulation does not replace reflection but allow detailed, context-sensitive analysis during process prior implementation. Finally, discuss inherently quantitative methods afforded simulations well discussions how can improve traditional forms thought experiments future-oriented assessment.