Mitigating Bias in Analytics: Data Collection Recap

We recently partnered with Tableau to create a virtual event series around mitigating bias throughout the analytics process. What better way to kick it off than with Renée Cummings!? Renée’s background is absolutely fascinating – she is a criminologist, criminal psychologist, AI ethicist, and data activist. She is also the founder of Urban AI where she helps organizations design inclusive AI, build responsible and trustworthy AI strategies, and develop and govern AI policies. 

Our conversation centered around the breakdown in systems to capture data that is truly representative and inclusive. If the data is not collected, or collected in a way that does not accurately reflect reality, any downstream system or analysis could not possibly consider or represent the individual left out. The type of impact this could have, and has had, is significant. Renée brought up examples in recidivism, policing, and substance abuse. How are we collecting and using data to provide resources to the people who need it? Even if these studies are correct and accurate, what are we doing with those conclusions?

Here are my top takeaways from our conversation:

  • It starts with education – we all need to understand where these systems might break down and how 
  • It is everyone’s responsibility to build ethical AI and we should start including questions about this when interviewing candidates
  • We need people at the table during the design discussions that represent those potentially impacted by AI to provide necessary context
  • Organizations need to start thinking about this early – do not wait until you are building hundreds of models to think about ethics and put policies in place
  • If your company is not doing this today, start the conversation – it should be ongoing and a part of the company culture

Be sure to check out the resources that Renée mentioned during the discussion below.


Additional References 


Our next event is Wednesday, August 19th, on the topic of design. We will cover the ways in which bias creeps into the design of analytical solutions and technology that prevents us from truly capturing data that is representative of its users. Stick around afterward to network live with attendees! 

Make sure to register here!

If you have any questions, you can contact our team at info@womeninanalytics.com.