RSM Seminar Series: Associate Professor Amy Dawel

Speaker: Associate Professor Amy Dawel

Title: Bias in AI-generated faces: Risks and opportunities

Date: Thursday, 8 October 2026

Time: 12:00pm – 2:00pm

Venue: Building 24, Copland, Room 1106, RSM Seminar Room

Alicia Gilchrist is the host of this visit

Abstract: 

Artificial intelligence (AI) can now generate faces so realistic they routinely deceive people, fuelling misinformation and fraud projected to cost USD 40 billion globally by 2027. Yet AI-generated faces are not identical to real ones. AI-faces differ systematically from human faces because the algorithms that produce them are biased—a feature that creates both risks and opportunities. A major risk is racial bias: AI often renders White faces as “hyperreal” while poorly representing people of colour, reinforcing inequity. The opportunity, however, is that these systematic flaws can be used as perceptual clues to train people to detect AI-faces. This talk provides an overview of current AI face technologies and what we know about the biases in them, highlighting associated real-world and scientific risks. I will then present a novel training method that leverages the biases in AI faces to train humans to detect them. Initial testing shows our training approach is highly effective, with all 45 participants improving from pre-to-post training and top performers approaching perfect accuracy. Critically, the method can adapt to new technologies by simply updating the training images. Overall, I argue for a human-centered approach to understanding AI faces and their biases, ensuring we keep people in the loop and develop explainable and accountable systems for AI detection.

Bio: 

Associate Professor Dr Amy Dawel is a clinical and cognitive psychologist at The Australian National University, where she leads the ANU Emotions and Faces Lab. Her research investigates emotion regulation and face perception, with a focus on how these processes support human connection and wellbeing. A current research priority involves understanding how people perceive and interact with AI-generated personas, including the detection of deepfake faces and the potential for human-machine collaboration to improve AI detection capabilities. This work has been featured in outlets such as The New York Times and National Geographic. Amy’s lab is also investigating relationships between AI chatbot use and children's wellbeing. Her research program is supported by the Australian Research Council and Australian Rotary Health, and has been recognised with the 2022 ACT Tall Poppy and 2023 McMichael Awards.