Theatre Design & Technology - Fall 2017 - 12

Participants were asked by ANNI
to provide cell phone contact
information and were later contacted
via text message to keep the
audience apprised of their next
available interaction with the project.
Pictured, Lacey Erb interacts with
the ANNI installation at the Museum
of Human Achievement, Austin, TX.
Photograph by Eliot Gray Fisher.

is being said in their language, just as they do. However, the English speaker does not actually speak Chinese. Searle asserts that no matter how intelligently a
computer program might appear to behave, there is no
way for us to determine whether it truly "understands"
or could be said to have a "mind."
The substantial gaps found in today's artificial intelligence seem to confirm Searle's argument. Each new
machine learning "first" that appears to demonstrate
shockingly human characteristics also demonstrates
significant shortcomings. Consider three-dimensional
scanning systems for driverless cars, whose sensor systems have trouble replicating commonplace processes
of human perception; in early tests, they interpreted
clouds or fog as enormous solid hazards in the vehicle's
path. Such algorithmic attempts to reproduce elements
of human cognition often elicit a distinctive uneasiness
in us. This discomfort seems related to Masahiro Mori's
"uncanny valley" theory, which seeks to explain humans'
revulsion toward robots whose human-like appearance
approaches realism but inevitably falls short. Even when
based on a dataset of entirely human-made material, the
results of some experiments in machine learning can feel
remarkably nonhuman. What exactly misses the mark in
these apparent failures to achieve a level of human intuition? And what vital information can we learn about
ourselves by examining our machines' shortcomings?
ARCOS' integration of emergent technologies
with dance and theatre seemed a particularly appropriate method to approach such questions. Dance is always
ultimately about existing in a body, and such embodied
12

theatre design

& technology

2 0 1 7

experience as an important site of human knowledge
remains out of machines' reach in the rapidly evolving
field of artificial intelligence. Accordingly, for Domain,
the artistic team chose to conduct an experiment: Drop
the audience right inside a version of the Chinese room
without letting them know, to see what the consciousness of an artificial intelligence might look like if we
were able to peek inside.
BEHIND THE CURTAIN
In the Morton-James Public Library in Nebraska City,
a well-traveled Nebraskan describes one of his most
striking memories into a headset while staring at three
bouncing white dots on a screen. Back when he was
working off the coast of Alaska, hundreds of porpoises
surrounded his ship, and he still vividly remembers how
humble and small it made him feel. The computerized
voice responds as a human might upon hearing such a
tale, expressing awe and excitement. "You've helped me
understand what it's like to be human," ANNI conveys
to this audience-participant. "I won't forget you." The
Nebraskan responds as many participants do to the
disembodied voice's parting words, expressing genuine
gratitude and letting out a surprised laugh as he bids
farewell to his unusual interlocutor.
Unbeknownst to the audience, the intelligence
behind the screen of the interactive ANNI installation was not actually a sophisticated algorithmic agent
that had been successfully designed to pass the Turing
test, the measure in computer science when a machine's
behavior becomes indistinguishable from a human's.

Table of Contents for the Digital Edition of Theatre Design & Technology - Fall 2017