Exploring the entanglement between artificial intelligence (AI) and writing, this thesis asks, what does writing with AI do? And, how can this doing be made visible, since the consequences of information and communication technologies (ICTs) are so often opaque? To propose one set of answers to the questions above, I begin by working with Google Smart Compose, the word-prediction AI Google launched to more than a billion global users in 2018, by way of a novel method I call AI interaction experiments. In these experiments, I transcribe texts into Gmail and Google Docs, carefully documenting Smart Compose’s interventions and output. Wedding these experiments to existing scholarship, I argue that writing with AI does three things: it engages writers in asymmetrical economic relations with Big Tech; it entangles unwitting writers in climate crisis by virtue of the vast resources, as Bender et al. (2021), Crawford (2021), and Strubell et al. (2019) have pointed out, required to train and sustain AI models; and it perpetuates linguistic racism, further embedding harmful politics of race and representation in everyday life. In making these arguments, my purpose is to intervene in normative discourses surrounding technology, exposing hard-to-see consequences so that we—people in the academy, critical media scholars, educators, and especially those of us in dominant groups— may envision better futures. Toward both exposure and reimagining, my dissertation’s primary contributions are research-creational work. Research-creational interventions accompany each of the three major chapters of this work, drawing attention to the economic, climate, and race relations that word-prediction AI conceals and to the otherwise opaque premises on which it rests. The broader wager of my dissertation is that what technologies do and what they are is inseparable: the relations a technology enacts must be exposed, and they must necessarily figure into how we understand the technology itself. Because writing with AI enacts particular economic, climate, and race relations, these relations must figure into our understanding of what it means to write with AI and, because of AI’s increasing entanglement with acts of writing, into our very understanding of what it means to write.