Stuart Russell has proposed that we should achieve AI alignment by building AI agents whichlearn our values or preferences from our behaviour, then select actions in accordance with them.One challenge to this approach is that different people value different things, but I set this challenge aside in order to focus on a second, which is that there is considerable complexity in the structure of...
Short Abstract In this paper, we will present an Ethics by Design approach for AI. Ethics by design approaches include ethical considerations in a systematic way in design processes by allowing ethical guidelines to be translated into concrete design practices. Our approach can be described in a five-layer model, going from abstract to concrete. It starts with (1) a total of six Ethics by...
Short abstract (See the attachment for the extended abstract)
Are there human activities that autonomous systems cannot perform? How does the answer to this question inform the design of autonomous systems? Evaluations of a technology or its features should be sensitive to the activities in which it is used. Each activity can be described by its overall goal, governing norms, and the...