Laura Weidinger

Laura Weidinger coauthored Using the Veil of Ignorance to align AI systems with principles of justice. The paper records equal contributions by Weidinger and Kevin McKee, and studies choices about principles governing an AI assistant.^[1]

She also coauthored Google DeepMind’s 2023 account of sociotechnical evaluation, which distinguishes model capabilities, human interaction and systemic impacts. These contributions connect Social Alignment with evaluation of AI in its social setting.

Reading her contributions together

Start with the principle-selection study and identify the population, decision and evidence it actually tests. Then use the sociotechnical evaluation account to ask what changes when an assistant interacts with people and institutions. A bounded preference result and a framework for organizing risks answer different questions; neither alone establishes legitimate choices for every affected community. The collective-life sequence in Key Papers: A Reading Guide develops that comparison.