Guest Seminar: "Making AI Intelligible; Designing for ethical and comprehensible intelligent systems"

Activity: Academic talk or presentation typesInvited talk


Abstract: The development of complex, intelligent systems has surpassed our ability to explain them. The market is increasingly populated with algorithmically-driven products for which users have limited mental models, putting their privacy at risk, and raising questions of human agency and control. Ensuring machine intelligence is ethically designed raises distinct challenges and implicates multiple spheres such as regulation, governance, education and design. In this talk I will present two studies; an investigation of the everyday use and comprehension of conversational agents, and a technique exploring privacy by design with system designers. I will also offer some early findings of a Wizard of Oz study testing player responses to an intelligent agent in Minecraft.
Period22 Feb 2017
Held atUCL