Ethics and Trust: Principles, Verification and Validation (Dagstuhl Seminar 19171)

摘要: This report documents the programme of, and outcomes from, Dagstuhl Seminar 19171 on "Ethics Trust: Principles, Verification Validation". We consider issues of ethics trust as crucial to future acceptance use autonomous systems. The development new classes systems, such medical robots, "driver-less" cars, assistive care robots has opened up questions how we can integrate truly systems into our society. Once a system is autonomous, i.e. learning from interactions, moving manipulating world are living in, making decisions by itself, must be certain that it will act in safe ethical way, able distinguish 'right' `wrong' make would expect it. In order for society accept these machines, also them, believe they reliable trying assist us, especially when engaged close human-robot interaction. seminar focused does with machines evolve, build `practical' trustworthy system, what societal implications. Key included: Change repair, AI decision makers, complex norms algorithmic bias, potential discrepancies between expectations capabilities machines. This workshop was follow-up 2016 16222 Engineering Moral Agents: From Human Morality Artificial Morality. When organizing this aimed bring together communities researchers moral philosophy artificial intelligence extend (social) robotics interaction research.