Accountable AI Systems 

The most recent advances of AI technology, namely neural networks, and their application to sophisticated pattern recognition tasks, such as image classification in automated vehicles, has led to a plethora of concerns regarding accountability, often couched in terms of the capability of these algorithms to explain their decisions. This presentation will address a different type of accountability: system accountability. We will look at the architecture of intelligent systems that are made of large numbers of intelligent agents and explore the issues and possible solutions to accountability when decisions and actions are the result of large numbers of individual decisions made by interacting intelligent agents.

☛ please register here