Abstract
Over the past seven decades, STEM-themed films have played a significant role in shaping public perceptions of scientists, mathematicians, and other technical professionals. This study investigates how such films encode social, racial, gender, and economic stereotypes in their portrayals of protagonists and examines how these representations evolve across time. Beyond film analysis, the project compares how these figures are described in human-written Wikipedia articles and in ChatGPT-generated summaries, exploring whether and how implicit biases persist, shift, or are reproduced across platforms often perceived as neutral or objective. Using a structured coding framework, the study analyzes narrative roles, character traits, markers of identity, and patterns of visibility or marginalization. By placing cinematic portrayals alongside digital knowledge representations, the project highlights how stereotypes may be reinforced, softened, or subtly transformed through algorithmic and crowd-sourced text production. From an educational perspective, this work demonstrates how mathematical and statistical analysis can be applied to questions of representation and bias. Engaging students in this research process invites them to critically examine how knowledge about STEM is constructed and circulated, while fostering a more inclusive understanding of who belongs in mathematics and related disciplines.