Prompt‐a‐Scientist: Using Generative AI to Disrupt Stereotypical Conceptions in Science Education

ABSTRACT

Stereotypical conceptions of scientists shape how students experience science, develop science identities, and make career choices. While instruments such as the Draw‐A‐Scientist Test (DAST) have been widely used to elicit these conceptions, fewer approaches focus on actively disrupting them. This paper introduces Prompt‐a‐Scientist, a pedagogical activity that uses generative Artificial Intelligence (AI) to support undergraduate students in recognizing and deconstructing their stereotypical views of scientists. In the activity, students first authored detailed narratives of a stereotype‐disrupting scientist and then translated these into prompts for AI image generation. Analysis of the resulting narratives, prompts, and images reveals a dual mediation process: students' conscious deconstruction efforts interacted with and were constrained by the AI's own default stereotypes. Results show that while AI can circumvent some clichés, it often reconstitutes others and struggles with nuanced identities like nonbinary gender. The subsequent collective reflection proved crucial, enabling students to problematize both the AI's biases and their own internalized assumptions. The activity complements rather than replaces existing instruments such as the DAST by shifting the pedagogical function from diagnostic assessment to a critical formative practice, offering a promising tool for both teaching and future research.