In 1950, Claude Shannon created Theseus, a maze-solving mechanical mouse that served as the world’s first practical demonstration of machine learning. Its brain was composed of electromechanical relays, like the same clicking metal switches used in this installation. Just one year later in 1951, Marvin Minsky built the first artificial neural network (SNARC) using vacuum tubes. These were the heavy, physical ancestors of the less visible components that are used today.

The progress of AI has since shifted from a steady walk to a dizzying sprint. This exhibit translates that exponential curve into light and sound. The relays begin with the slow, deliberate cadence of the 1950s, gradually accelerating into a frantic mechanical blur, mimicking the shortening timelines felt by many today.

Every cycle, the machine chooses a random year that AGI will arrive. These years are pulled from the real-world predictions of industry leaders. Some believe we reached AGI months ago, while the majority believe it is just a few years away. When the year is finally reached, we will be at the finish line of a race that began with a single clicking relay in 1950.

This is a kinetic art piece I completed in March 2026.