The Value Alignment Problem in AI: How to Use Ethics to Govern AI Artefacts

The ultimate purpose of AI is the design and construction of autonomous and adaptive artefacts. Artefacts perceive their environment and can decide how to act without the direct intervention of humans. The key challenge of the field is how to govern that autonomy. One approach is to imbue values into such artefacts so that their behaviour is objectively aligned with human values. This talk articulates the challenge and explores how values can be engineered into artificially autonomous systems.