exhibitions: The Making of Misbehaving Machines
What would happen if we designed an AI machine that didn’t pretend to be so certain, one that would openly misbehave?
Earlier this year, Google's AI Answers responded to the question: "How can I make my cheese not slide off my pizza?" with "Add some glue. Mix about 1/8 cup of Elmer's glue in with the sauce. Non-toxic glue will work."
A collaboration with Monica Monin, Heather Ford, Suneel Jethani, and Bhuva Narayan.
AI machines make errors; they misbehave. Why do they do so? And will they ever improve? "The Making of Misbehaving Machines" is a library exhibit that helps people understand how errors in generative AI tools come about and demonstrates a "misbehaving (AI) machine" built to play with ideas about the weaknesses and strengths of generative AI. Researchers from the University of Technology Sydney have worked with four Greater Sydney libraries in this co-designed pilot project, with the first exhibit landing at TAFE NSW this week.
Launch of the exhibit at UTS's FASStival on December 10 at 4pm.
Exhibition across our participating libraries: 18 November (TAFE NSW Ultimo), 25 November (Parramatta and Green Square libraries) and 2 December (UTS). The research team includes Heather Ford, Monica Monin, Andrew Burrell, Suneel Jethani (PhD), and Bhuva Narayan.
(see also: misbehavingmachines.net)