Publikation
CHIM - Chatbot in the Museum: Exploring and Explaining Museum Objects with Speech-Based AI
Oliver Gustke; Stefan Schaffer; Aaron Ruß
In: Sonja Thiel; Johannes C. Bernhardt. AI in Museums: Reflections, Perspectives and Applications. Pages 257-264, ISBN 9783839467107, transcript Verlag, Bielefeld, 2023.
Zusammenfassung
CHIM - Chatbot in the Museum is a research project that kicked off with a first brainstorming meeting in January 2020 and ended in June 2022 with the evaluation of the field test data collected with a prototype chatbot application at the Städel Museum,Frankfurt am Main. CHIM will be available as opensource software on GitHub in the second half of 2023. CHIM explores the use of AI-supported, speech-based interactional conversation systems in educational community work in museums. Our aim is to give standard media guides a voice (and a brain). CHIM is a prototype intended to outline a preliminary stage of next-level digital museum guides, which communicate content about museum objects not only ‘one way’, but instead bi-directionally. CHIM is a step towards future forms of more participatory approaches to communicating with visitors.
The idea for this project was born in a workshop on education in the museum and outreach strategies. One of the participants stated that she normally would not book or use any personal audio guides or media guide systems, because she loves simply browsing among the objects. But then, in front of particular objects, questions sometimes do arise. Questions that are often so specific that she is not able to answer them with a simple, short internet search. That, one might say, ‘use story’ led to the idea to create a chatbot application for mobile devices that refers to a very specialized database and tries to detect users’ intentions in order to find accurate answers: CHIM-Chatbot in the Museum. We began by specifying 15 (in the end, we conducted testing with 13) artworks at the Städel Museum, Frankfurt am Main, as ‘test objects’. We then compiled content and designed a demonstration and test app.