Unconscious bias in ML
Priming and anchoring in Data Science
Introduction
During the last weeks, I’ve been listening on Audible to the book “Thinking, Fast and Slow”, a masterpiece of Daniel Kahneman, Nobel prize for economic science, where the author presents the two systems that drive the way we think and how they work. These two systems are introduced to us as if they were two characters of a story, with their own personality and intrinsic characteristics, interacting between them as people do. In this context, the author sets the field for us to recognize in ourselves a System 1 that’s fast, intuitive, and emotional, and a System 2 that’s slower, but also more deliberative and logical. Following these two characters, Kahneman takes us through a journey where the interaction in between them helps us answer questions such as why is there more chance we’ll believe something if it’s in written in bold? Why are judges more likely to deny parole before lunch? Or why do we assume a good-looking person will be more competent? Moreover, the author spends a lot of time talking about how the way we think impacts activities such as sampling and probability thinking.
These last two points got me thinking, since, in Data Science, the precision and accuracy of our sampling, as well as the lack of bias in our beliefs and targets, are two key points for…
