sponsored by the Partnership on AI

Shared between the following two papers:

Transparency and Explanation in Deep Reinforcement Learning Neural Networks

Rahul Iyer, Yuezhang Li, Huao Li, Michael Lewis, Ramitha Sundar, Katia Sycara.

For AI systems to be accepted and trusted, the users should be able to understand the reasoning process of the system and to form coherent explanations of the systems decisions and actions. This paper presents a novel and general method to provide a visualisation of internal states of deep reinforcement learning models, thus enabling the formation of explanations that are intelligible to humans.

An AI Race: Rhetoric and Risks

Stephen Cave, Seán S ÓhÉigeartaigh

The rhetoric of the race for strategic advantage is increasingly being used with regard to the development of AI. This paper assesses the potential risks of the AI race narrative, explores the role of the research community in responding to these risks, and discusses alternative ways to develop AI in a collaborative and responsible way.