Labor as a Human Product
Through the rise of AI as a modern-day tool, we are seeing an mechanization of all kinds of labor; however, the argument as to whether this development is beneficial is hotly debated. As numerous corporations have begun to utilize AI as a tool to optimize the means by which products are created—lines of software code, poetry, and pixelated images—even the human nature of art as labor is an idea that is slowly beginning to die out.
When one thinks “labor,” the first thing to come to mind is usually a person doing some sort of job, and this includes artistic labor. In response to the collapse of 1930s markets, the United States government recognized that artists were workers whose skills qualified as a form of labor and employed them through the Federal Art Project. The products of their labor are personal and human and express the troubling times, displaying a variety of emotionally complex topics such as adversity and hardship. AI, as it exists today, can generate sophisticated images, and as an extension allows anybody to create one within seconds. In doing this, however, our embodied human touch to the artwork is lost; we struggle with terms like “uncanny valley” or “slop” to describe the new relationship.
Let’s look more closely at a lithograph by Bernard titled Sheep Shearing. We might notice that the human figures are somewhat stylized, in ways that differ from his other works at the New Deal Museum. Was his style influenced by his friend and housemate, Jackson Pollock? Is there some underlying irony to such a pastoral scene created during economic and ecological stress? The pattern of a sheep’s trimmed fleece is echoed in the texture of a tree’s leaves, above, but not quite. When we look closely, all of Schardt’s individual strokes are visible. How would these and other features enable an AI program to generate an image described as “Sheep Shearing New Deal lithograph in the style of Bernard P. Schardt”? In fact, AI programs are “trained” on massive data sets of images already created by others; mathematical models identify patterns based upon pixels and coordinates. Here are just a few lines out of hundreds comprising the lamb’s face at the print’s right side, using RGB (red-green-blue) numbers:
150 144 122 158 152 130 151 143 122 143 135 114 144 136 113 147 139 116 158 147 125 169 159 134 158 148 123 149 139 114 161 151 126 166 156 131 161 153 130 169 163 139 173 167 145 171 166 144 165 158 139 160 153 134 163 156 137 166 159 140 164 159 139 168 163 143 168 163 144 159 154 135 153 148 129 165 160 141 170 165
AI aggregates massive amounts of the human product; what it produces is something utterly different than artistic labor, even when its algorithms introduce random strokes into a visual composition. Unlike works made under the FAP, there is no purpose behind AI works, since it is simply the product of pre-existing art documented online, being mashed together to fulfill a specific prompt.
Works Consulted
-- Boxentriq. "Pixel Values Extractor." Link
-- Isaacs-Thomas, Bella. "How AI Turns Text Into Images." PBS News 12 Jan. 2023. Link