Runaway Models is an interactive story that explains all the stages where human biases can be unintentionally learned, propagated and amplified by Language Models, and explores how these biases can be mitigated. This interactive story is designed for a lay audience with a limited to working knowledge of machine learning and AI technologies.
Runaway Models is a dissertation project for the Design Informatics programme at the University of Edinburgh, designed and developed by Tamara Lottering under the supervision of Dr Benjamin Bach (University of Edinburgh) and Mennatallah El-Assady (University of Konstanz).
This project is a work-in-progress. Please contact me at lotteringtamara@gmail.com for any thoughts, questions or requests for collaboration!
Explore Runaway Models