Should we teach a machine to worry for us?
Ecology of Worries is an animation featuring variously evolved critters that are driven to speak by machine learning algorithms trained on actual human worries. The creatures’ performance of the worries spans a gradient of intelligibility, reflecting on the evolution of machine learning systems and whether or not we should teach a machine to worry for us. The animated characters are also representations of our collective worries, given life through evolving algorithms.
The Ecology of Worries animations were projection mapped to buildings at PASEO Project in Taos, NM, and ILLUMINUS in Boston, MA. For these festivals, the worry critters were integrated into the built environments creating an opportunity to reflect on the physical burden of anxieties we carry with us day to day. The audio was played through speakers in the street and also available via headphones. For an exhibition at the Brookline Arts Center in Brookline, MA we created a series of posters with images of the critters which could be activated using augmented reality (AR). Additionally, we have exhibited a single channel version of the animation. For the PASEO Project, we created a site-specific critter using worries recorded by people in Taos and imagery inspired by ecology in New Mexico. For the Brookline Arts Center, we created a site-specific critter prompted by written worries from children attending classes at the center and imagery inspired by ecology in Massachusetts.
The critters are trained using an archive of over a thousand recorded worries from people in the US and abroad which we began collecting in 2016. Some of these worry critters are driven by a TextGenRnn neural network trained on our collected worries archive. Others are driven to worry by the Generative Pretrained Transformer 2 (GPT-2), which was originally dubbed by commentators as the AI that was too dangerous to release (but it was released anyway). The critters created for Taos and Brookline used GPT-3. The resulting manifestations of the algorithms are presented as a hand-drawn audiovisual menagerie.
This animation features excerpts from six critters generated with SampleRnn and GPT-2.
Our illustrations are inspired by insects and plants. Derived from an ecosystem we are dependent on, while simultaneously providing imagery that even when personified is seemingly neutral and non-specific. By characterizing the synthetic worries of various sophistication as variously evolved creatures we are engaging the empathy of the viewers. We’re crowdsourcing worries from people, feeding them into machines, getting synthetic worries out, then feeding them back out to the people, who will produce more worries, etc. The critters in Ecology of Worries appear alive due to their very real flaws. The project defamiliarizes the peppy digital assistant voice by training these creatures to worry using our communal woes. The creatures become uncanny through a juxtaposition of familiar and abstract concerns.
The Eliza effect, discovered in the late 60s, led people to perceive a chatbot as intelligent and worth confessing to. Ecology of Worries flips the dynamic to have the machines confess to us and put us in a thoughtful and sometimes hilarious state of mind. The overall experience creates space to explore an evolving cultural ecology of empathy via a hybrid, human-machine ecology. Can we learn about each other this way, cut to the gist of what’s underneath?
We continue to collect worries and the public is invited to record worries anonymously at https://worries.io.