What If Classroom AI Treats Some Students Worse Than Others?

In this episode, we examine a troubling MIT study showing that AI chatbots can respond differently based on who they think they’re talking to, sometimes giving less accurate, more patronizing, or even mocking responses to users described as less educated or non-native English speakers.This episode looks at:what AI bias can look like in practice why district leaders need better ways to evaluate these toolshow researchers and schools might work together on accountability why AI alignment is not a finished problem what it means when AI systems “get to know” users over time