Executive Summary - Accountable AI

The LCO’s Accountable AI paper considers how to ensure legal accountability when governments and public agencies use AI to make or assist decision-making in the civil and administrative justice systems.1 The context for this analysis is the extraordinary growth in the use of AI and ADM by governments across the world. [...] The breadth and pace of government AI systems reflects the perceived potential of AI to improve the accuracy, speed, and consistency of government decision-making. [...] Earlier papers address the use of AI and ADM in the criminal justice system,2 government AI systems development,3 how to regulate AI,4 and the use of AI to generate evidence in criminal proceedings.5 This report is part of the LCO’s ongoing AI, ADM, and the Justice System project. [...] potential to significantly improve the administrative state.8 Risks and Harms of AI Governments around the world are using AI Governments must respond to the well- across a broad range of areas, services, and documented risks and harms of AI and ADM functions. [...] The provincial government, community legal clinics, members of the private bar, academics, the judiciary, tribunals, and technologists should come together to consider the following issues: • How to determine the evidence required when assessing a government AI system? • How to assess the “reasons” of a government AI system? • How to assess the risk and impact of a decision made or aided by a gove.