What do we know about our error rates? We probably expect to get some level of errors from the AI, but what do we know about the error rates in what we’re putting into the AI?
So I’ve been doing a series of workshops and blog posts on error rates and data quality, and was especially pleased to have an opportunity to do a Rosenverse Live webinar for the Rosenfeld community.
My takeaways are:
- There are many ways in which the data we collect isn’t entirely accurate
- To track your error rates:
- Count the errors
- Think hard about what you divide by
- There are many ways in which the data we use isn’t entirely accurate
- Data quality is important but not easy
- AI may help to start useful conversations about error rates and data quality.
Thank you to Lou Rosenfeld for giving me the opportunity to talk about error rates in the AI age, and extra thanks to everyone who came to the webinar and contributed their thoughts.
- Watch the video recording of Rosenverse Live: Garbage in, garbage out – measuring error rates and data quality to get ready for AI (needs free account)
- Download the slides: (pdf) Garbage in, garbage out – measuring error rates and data quality to get ready for AI
- Read about six ways to think about errors
- Find other webinars and research in this series about error rates
- View the webinar slides:
View the slides as a PDF: Garbage in garbage out for the 2026 Rosenverse