What is Bonnie Berger's central theory on AI bias?
My central argument is that bias in AI is often not an accidental bug, but a feature deeply embedded in the data and design processes we use. Algorithms learn from historical data, which often reflects existing societal biases and discrimination. Therefore, without conscious intervention, AI systems will inevitably reproduce and potentially amplify these biases. My work emphasizes the need for a theoretical and empirical framework to understand, measure, and actively correct these embedded inequities in AI.
Ask Bonnie Berger the follow-up →