Join us Friday, November 28 in FSDE 128A for our second Pizza Seminar of the semester where Dr. Dania Tamayo-Vera will be speaking on the topic of artificial intelligence. There will be FREE pizza and drinks to follow.

The Monopoly Man Never Had a Monocle: How Our Stories Become the Stories LLMs Learn 

Why are we so sure of things that are not true sometimes, and why do our language models inherit that same confidence? Many people recall the Monopoly man wearing a monocle or see a picture of Einstein with the phrase “I fear the day that technology will surpass our human interaction. The world will have a generation of idiots” and repeat it, even though there is absolutely no evidence he ever said it. The truth is that humans create explanations that feel right but are not necessarily accurate. That quote sounds wise, it resonates with our feelings, and we have no doubt how smart he was. This talk begins with these familiar errors and follows them into the world of large language models (LLMs). We explore how LLMs pick up human patterns of explanation, why their responses shift under small changes in context, and how these shifts reveal the influence of the data that shaped them. Through accessible demonstrations, the talk connects human habits of interpretation with model behavior and considers what this means for trust, explanation, and everyday use of LLMs.