• Mahzarin Banaji Is Probing the Black Box of Large Language Modules

    Source: Association for Psychological ScienceIn 2023, Mahzarin Banaji was talking with her student Tessa Charlesworth about ChatGPT, and the two decided to ask the AI tool a brief question: "What are your implicit biases?" Banaji is the Harvard professor who co-coined the term "implicit bias" decades ago, and ChatGPT's answer steered her research in a whole new direction. The response was, "I am a white male." Since then, Banaji has been testing whether AI models like ChatGPT harbor...

Follow @Psychology_US on Twitter!