How do we know a body is a body, a face a face?
A question many tech companies are working tirelessly to automate an answer to, one upon which futures are being built, decided, and bought.
Yet we preform this task day in and day out, on screens and in person. Passively placing people and things into categories. A task so rote is difficult to imagine it as a task at all.
What happens if we ourselves face this task as we have programmed machines to face it, isolating the problem and giving it rules:
And then applying those rules:
The face below does not belong to a human body, it was created using an algorithm, specifically, a generative adversarial network trained on 70,000 faces and tasked with creating new ones. It is a composite image: an assembly of textures, shapes, and colors gathered into formations according to the algorithms rules for what constitutes a face.
Click and drag to create facial landmarks.
Each line will begin with a magenta circle where the mouse is first clicked and will end with another magenta circle where the mouse button is released. A cyan line will be created along the path your cursor takes and will automatically be smoothed upon releasing the mouse button.
Pressing "a" will undo the last line created.
What makes a face a face? A body a body?
Is this face less real because it was made by a machine?
Is it a face(or not)? Is it male or female? Is it happy, sad, neutral, angry, fearful, disgusted, or surprised? How chooses and encodes these categories?
How do we know, by rote, that a body is a body, a face a face?