|
Abstract:
|
Artificial intelligence (AI) and machine learning (ML) are technical terms used loosely and interchangeably by consumers of statistical science. They say that AI is used by executives in the boardroom, while ML is preferred by technicians. In applied statistics, oftentimes as consultants, we might struggle to temper expectations levied on our work, in particular when our clients do not grasp uncertainty and bias and instead view AI/ML methods as an end, not a means. Over lunch, let's ponder our roles as ethical practitioners in this era of ever-increasing pressure in the AI/ML setting to deliver results that our clients continue to expect. We draw on specific examples of AI/ML applications in health policy, focusing on algorithms to predict outcomes using administrative claims and electronic health record data. Together, we develop a framework to "sell the story" in a way that balances scientific and practical perspectives on the ethical application of statistical methods to solve pressing problems in health and social policy.
|
