AI Writing Is Always Embodied: Building a Critical Awareness of the Invisible Labor of Humans-in-the-Loop in AI Products

I argue that composition studies must build critical awareness about how humans from the Global South train AI with their writing embodiments. To draw our attention to how those working in the Global South train AI in harmful conditions, even though AI companies use algorithms and terms of service to smooth away these embodiments, I adapt the concept of humans-in-the-loop. Critical awareness of humans-in-the-loop moves scholarship in writing studies from a focus on AI-human collaboration that begins after an AI produces a text to one that requires us to see how AI products are always already human authored. Through a case study of Google Translate, I demonstrate how a critical awareness of how AI can erase the writing embodiment of humans-in-the-loop affords me opportunities to ask generative questions: How does language translation play a role in the erasure of embodied writing? Why does AI produce with bias toward marginalized populations when marginalized populations are those that moderate AI? Overall, I ask compositionists to see AI products as already human authored so that writing studies can consider the invisible labor of humans-in-the-loop as the field moves forward in researching AI.