Rapid Trust Calibration through Interpretable and Uncertainty-Aware AI
Published: 10 July 2020
Authors: Richard Tomsett, Alun Preece, Dave Braines, Federico Cerutti, Supriyo Chakraborty, Mani B. Srivastava, Gavin Pearson, Lance M. Kaplan
The paper starts from a practical question: in high-stakes settings, an AI prediction is not enough if operators cannot see how reliable it is.
The proposed idea of rapid trust calibration is to give people enough evidence to decide quickly whether to trust the system, how much, and where uncertainty is still material.