Be Prepared to Show Your Working! - Professor Sir David Spiegelhalter

Explore the complexities of algorithmic decision-making and its impact on society in this thought-provoking lecture by Professor Sir David Spiegelhalter. Delve into the importance of transparency, validation, and explainability in automated advice systems, tracing the evolution of these concerns from the 1980s to present day. Learn about the concept of 'intelligent transparency' and discover a framework for evaluating algorithmic trustworthiness. Examine real-world applications through the Predict system for breast cancer treatment decisions, featuring multiple levels of explanation. Gain insights into the challenges of communicating uncertainty, ensuring fairness, and maintaining transparency in proprietary algorithms. Engage with critical questions about the role of algorithms in shaping our world and the ethical considerations surrounding their use in sensitive decision-making processes.

Introduction
Summary
Trust
Trustworthy claims
Structured claims
Algorithm competition
North Devon
Bootstrap
Probability distribution
Carl Dahle
The Turing Test
Designing Phase 3 Randomized Trials
Designing a Diagnostic System
Transparency
Interpretability
Global explained ability
Local explained ability
Black boxes
Predict
Predict Version 21
Survival Curve
Information
Download
What a shame
Algorithms are proprietary
Compass
Cynthia Rudin
Alan Turing
Communicating uncertainty
Probabilities
Test of calibration
Uncertainty
Fairness
Emails
Conclusions
Questions