Diversity, Inclusion and Bias in AI Systems
Most companies are increasingly trying to leverage data science to achieve their business goals, but it's easy to create biased models that disadvantage underrepresented groups. In this episode I get to chat with Labhesh Patel, CTO and Chief Scientist at Jumio.
We start by looking at common types of data bias around representation and features that may be highly correlated with protected features (zip codes that are highly correlated with race, for example) and the challenges of supervised learning where the team tagging your data can also introduce bias into models.
We then discuss the benefits of increasing the diversity and inclusiveness of both your tagging and modeling teams and questions you can ask as an engineering leader to make sure your data science team is being thoughtful about the potential impact of the models they're building.
VIDEOS RELATED TO MANAGING
Brandon Turner, Senior Director at Rapid7
Rob Zuber, CTO at CircleCI
Jerrold Jackson, Head of Machine Learning & Data at EXOS
Tim Olshansky, at
Camille Fournier, Head of Platform Engineering at Two Sigma
Russell Smith, CTO at Rainforest QA
Randy Shoup, VP Engineering and Chief Architect at eBay
Heidi Waterhouse, Transformation Advocate at LaunchDarkly
Mai Irie, Director of Engineering at Spring Health
Johnny ray Austin, CTO at Till
Juan pablo Buriticá, Head of Engineering, LATAM at Stripe
Jeff Smith, Senior Research Engineering Manager at Facebook Artificial Intelligence Research (FAIR)
Copyright © 2022 CTO Connection, All Rights Reserved