Abstract

Science museums are under pressure to present something about AI. The usual response is an exhibition of the newest tools. That exhibition is out of date before it opens, and it often leaves visitors worried or impressed but no better informed.

For Adventures in AI, the Exploratorium took a different approach. We built hands-on exhibits about the parts of AI that do not change: how a system learns from examples, where training data comes from, what a model cannot perceive, and why a confident answer is not always a correct one. These ideas will still be true when today's tools are gone.

This presentation tours the exhibits we built, including the ones that did not work. It describes how we designed for the fear and the excitement visitors bring with them, without pushing them toward either. It shares the decisions that mattered most: exhibits that make disagreement visible, activities that ask visitors to state an opinion, and experiences that let a group build a
shared understanding of how a system works.

In the AI era, the most valuable thing a science museum offers is not access to the technology. It is a place to think about it with other people.

Biography

· 2012–Present, Manager, New Media Exhibit Development, Exploratorium
· 2022–Present, Gallery Curator, Human Phenomena Gallery, Exploratorium
· 2025–2028, Lead Developer, SAIL (Supporting AI Literacy through Exhibits and Educators), IMLS
· National Leadership Grant with the Museum of Science, Boston
· 2024–2027, Project Director, EDGE of AI, IMLS Grant, Exploratorium
· 2020–2023, Lead Exhibit Developer, Kiewit Luminarium, Omaha, Exploratorium
· 2019–2022, Lead Exhibit Developer, The Scientific Center Kuwait, Exploratorium