Sven Nyholm: Responsibility Gaps, Value Alignment, and Meaningful Human Control over AI

  • 🎬 Video
  • ℹ️ Published 2 years ago

Sven Nyholm (Utrecht University) presents his talk on responsibility gaps, value alignment and meaningful human control over AI.

I will first distinguish among four different kinds of responsibility gaps, and briefly compare those to four kinds of responsibility gaps identified by Filippo Santoni de Sio and Giulio Mecacci. Next, I will compare what I call forward-looking positive responsibility gaps to the so-called value alignment problem in AI ethics. Lastly, I will compare two attempts to fill responsibility gaps: the solution I have argued for in previous work and what Santoni de Sio and Mecacci argue for in their work. Both solutions can be viewed as attempts to specify what it is to have meaningful human control over AI. Both attempted solutions, I will argue, face important problems.