I believe any system that's using algorithms to make automated decisions should get it called like a hygiene score, kind of like when you go into a restaurant. I have seen work out there by data for Black Lives and other policy groups that advocate for nutrition labels for algorithms. And then I've also seen model cards. Model cards were something that were developed by Meg Mitchell and Timmy Cabrue when they were at Google before the stochastic parrots debacle.
Meredith Broussard is one of the most visible, vocal leaders in the emerging field of algorithmic accountability. Professor Broussard is a data scientist and Associate Professor at NYU whose research focuses on AI in investigative reporting and using data analysis for social good. Meredith is the author of Artificial Unintelligence: How Computers Misunderstand the World and the forthcoming More Than a Glitch: Confronting Race, Gender, and Ability Bias in Tech.
Among other things, Professor Broussard was featured in the seminal documentary Coded Bias. Today's discussion is about one of the most important topics in our field and in this episode we explore it with someone whose name is on a shortlist of AI ethics pioneers. You’ve heard me say repeatedly coursework in AI ethics should be required for every student graduating with a technical degree. Here's why!
Listen and learn...
- How AI reveals bias encoded in society
- Why it's important to always ask "what could go wrong"
- What is the new field of "algorithmic accountability reporting"
- What the Blueprint for an AI Bill of Rights may mean for bad acting companies
- What's the right role for the federal government in AI regulation
- How to assign an "accountability score" to algorithms
- The ethical issues related to AI we'll be discussing in a decade
References in this episode...