posted on 2022-03-28, 16:31authored byCameron Edmond
Machine writing is defined in this thesis as the explicit use of computational methods to compose literature. The form has seen a surge in popularity as part of academic and journalistic conversations regarding the role of artificial intelligence (AI) in creative practices. Currently, creative coders, literary scholars and computer scientists are grappling with what does and does not constitute AI-penned literature, occasionally leading into conversations regarding singularity-type events. This thesis seeks to ground the conversation, focusing on past and present machine writing and evaluating how it relates to the idea of AI literature. To do this, the common theme of comparing literature by machines to literature by humans will be challenged. Through close readings of both output and code, I argue that to understand AI literature, we must explore the poetics and consequences unique to machine writing.
In order to highlight changes in the form, this thesis will develop a history of machine writing. Trajectories from the language games of the OuLiPo group and the Beat Generation's cut-up experiments to Twitterbots and algorithmically generated novels will be charted. Through analysis that focuses on both the literary output of these machines and their constraints/codes, the distinctly inhuman qualities of machine writing will be explored. This analysis will reveal the capacity of machine writing to transform texts, create unthinkable worlds, and mirror the anxieties of the information age. In short, machine writing will be analysed on the basis of its own literary and linguistic techniques, and the effects these techniques produce, establishing a sort of "machine poetics".
Uncovering the poetics of these machines then reveals several other concerns. Authorship becomes complicated by the use of algorithms, corpora and other authorial forces. Further, the utility of machine writing to both promulgate and dismantle oppressive structures through unthinking automation is revealed. In exploring these topics, this thesis maps the current climate of machine writing literature and provides insight into what may lie beyond the AI literature frontier.
History
Table of Contents
Introduction : machine writing -- Chapter 1. Historical machines : charting and contextualising machine writing trajectories -- Chapter 2. Collisions of prose and code : adaptation and analysis through machine writing and reading -- Chapter 3. Nauseous machines : subversion through absence in machine writing -- Chapter 4. Noisy machines : the use of noise in machine writing -- Chapter 5. Cyborgs & tarantulas : cut-up, corpora and challenges to authorship -- Chapter 6. Weaponised machines : forging and dismantling oppression through machine writing -- Conclusion : the AI literature frontier -- Reference list.
Notes
Theoretical thesis.
Bibliography: pages 225-266
Awarding Institution
Macquarie University
Degree Type
Thesis PhD
Degree
PhD, Macquarie University, Faculty of Arts, Department of English