If You Can Read This, You Are Not A Machine

[Update 13/05/2024: as of version 4o, released on this date, ChatGPT is able to accurately extract text set in Adversarial Mono from an image.]

The act of extending our selves with technologies we create has been fundamental in the trajectory of humans. We do not simply use our tools: as we create them, they also create us, modify us, shape us.
I'm interested in the nature of this entanglement we are in with our machines, this relationship: the cyborg condition. This condition, while not new, is certainly accelerating, and I find myself struggling sometimes to still sense a boundary between myself and my machines.

01 / Adversarial Mono

The first part of this project was a way of reasserting this distance: I developed a working font which humans can read, but machines cannot.

02 / Feelings About Machines

In the second part, I gathered seven short texts I'd written describing personal and visceral emotions about my relationship with machines into a zine. The zine was set in Adversarial Mono, printed in risograph in 25 copies and hand-bound.

There is no information in the zine that needs to be hidden from the machines (in any interpretation of the idea). Why, then, so many layers of removal? A font they can't read, a text that devolves towards almost body horror, a printed object that cannot become training data, a tactility they cannot register? I came to realize this project isn't so much about shutting the machines out as about trying to find a sense of allyship with the other humans.

Presentation

With the zine in their hands, I asked the audience to turn to each other, and to each choose a text for the other to read to them. Finally, I invited them to contribute to an open document (projected overhead for the duration of the exhibition) and leave their own message, for human eyes only.

Process & Research

[In a study by Dr. Tamara Makin at UCL’s Institute of Cognitive Neuroscience, the researcher] "invited participants to wear a robotic 'Third Thumb' (...) Less than a week after the introduction of the new finger, fMRI scans of the participants’ brains revealed that their natural body representation—the “mental picture” of their own hands in their motor cortex—had changed. (...) Their fingers were collapsing into one another, as though they were being squeezed together to make room for the new thumb."

[01]: An Afghan war rug (2010) depicting the 9/11 attack, with three towers instead of two. [02]: Typical image of a hand generated by Midjourney (2024), with extra fingers. I began my research into the possibility of an adversarial font from the question: what are "machine-like" mistakes, and what are "human-like" ones? What do the types of mistakes each of us makes reveal about "human-ness" and "machine-ness"?

From "The Invisible Hand" (2021), by Claire L. Evans, published on pioneerworks.org

Google Lens attempts (and fails) to read text set in Adversarial Mono