
Research programs: Computational Communication & Culture, Digital Publics
Abstract: This thesis seeks to understand how imitation works as an integral function of Generative AI technologies and their machine learning architectures. I use ‘simulation’ and ‘mimesis’ as my guiding conceptual tools, and specifically use two often overlooked tensions that are present in each respective body of literature. In the case of simulation I draw on the paradox that originates in Plato’s Cratylus that: if a copy comes too close to an original, it is no longer a copy, but another original. While in the case of mimesis, I draw on the idea that the representation of something (e.g., a photo of a group of people) can influence how the thing that is being represented is understood and viewed by others (e.g., the group of people that the photo is of). I use these two tensions to guide an investigation into the function of machine learning architectures, the design of interfaces of GenAI technologies, and to understand the wider ideological drive to build a machine that simulates the human faculties perfectly (often referred to as Artificial General Intelligence). In doing so, I use the two tensions described above as vehicles to examine issues of power latent in the automation of communication.
Principal Supervisor
Associate Supervisors