Shannon’s information theory, AI, and fiction

This essay is speculative. Rather than proceeding through a mathematical understanding of the potentials of AI, particularly as it concerns its capacity to mimic works of literature, it proceeds through readings of a key article related to the emergence of AI, or the groundwork that proceeded it: Claude Shannon’s 1948 essay that founded Information Theory ‘A Mathematical Theory of Communication’. I argue that given that this essay establishes the ground or foundations for Information Theory, that it sets crucial parameters through which work that depends on the processing and transmission of information in engineering terms continues to apply to its functioning. Using this as a way in, I look for moments that allow insights into differences that might be perceived in human understanding and mathematical processes that seek to mimic human understanding such as LLMs and Generative AI. Because of its speculative nature, parts of this paper will outline problems, and point to them as areas that need to be examined rather than fully exploring them. I then extend this to a brief reading of Alexis Wright’s Praiseworthy as an example text.