The Machine in the High Castle: A Critical Approach to Content Generators through the Rhetoric of Sci-fi, Gen AI, and Writing as Labour

A considerable portion of the discussion surrounding generative artificial intelligence (AI), large language models (LLMs), and text generators in the humanities has revolved around the tension between academic integrity and good pedagogy. Namely, instructors and administrators have spent the last few years scrambling to consider how we might prevent students from cheating and teach students to use a technology that we are only just beginning to understand ourselves. As an instructor, I understand that these are important concerns that warrant our attention. As someone who studies media and literature, I am also very interested in how creators will implement these tools in their art in new and interesting ways, even as I am concerned how corporate interests might exploit these same tools to further commodify mainstream entertainment and commercial art. In this article, I focus on the latter concerns with the view that, as a teacher-scholar, a better comprehension of how recent technological advances (as well as our collective perceptions of those advances) are affecting the form and content of creative expression will ideally lead to a fuller understanding of how we might engage with those same advances in the classroom. How we frame and imagine technologies matter: a recent study by Pataranutaporn et al. demonstrate that the implicit biases we carry into human-AI interaction influence its effectiveness (1077; 1083). Artists are responding to this understanding in literature and art, and initiatives such as the Future of Life Institute’s World Building Contest ask competitors to envision and design sci-fi futures that imagine generative AI as a tool that works towards humanity’s collective good. These efforts are in response to a largely dystopian treatment of AI that persists in sci-fi literature and the popular imagination. For better or worse, we have been quick to associate generative AI and LLMs with some of its most extreme literary counterparts. At the same time, a healthy skepticism of technological developments like text generators (e.g., ChatGPT, Gemini, CoPilot, etc.) and their commercial applications can help consumers anticipate the dangers and downsides of these tools. Scientists in both industry and the academy find themselves contending with a public perception largely formed by the likes of Ridley Scott’s Terminator and Stanley Kubrick’s HAL when advocating for this technology (Zador & Lecun), and scholars such as Theresa Heffernan see the need for a more focused discourse surrounding these technologies if we are to have a productive conversation without merely letting tech companies adapt tropes from utopian science fiction into their own advertisements and propaganda. Rampaging robots and murderous operating systems may be hyperbolic expressions of cultural anxieties related to our own identity as humans, but the popularization of chatbots at the advent of ChatGPT has also brought into focus legitimate concerns about jobs and labour, especially in relation to the generation of creative content. Even in this much narrower subfield of text and image generators, science fiction once again predicts and projects anxieties about technology’s misuse. Yet writers’ imaginations have tended to be more restrained in their pessimism and more focused in their concerns regarding the commercial mechanization of their own labour; models of text- and art-generating inventions in the works of George Orwell, Philip K. Dick, and others often tackle this subject with irreverent satire or a cynical eye towards corporate greed, and the automation of activities that we usually consider creative acts, especially in entertainment industries, reduces and commodifies artistic expression as exploitable labour. I begin this essay with a critical reflection on how certain companies have already started enacting bleak futures that surpass the cynical assumptions of writers like Orwell and J. G. Ballard …