Abstract
Increasingly, people engage with Large Language Models not as tools for information retrieval but as companions, confidants, and therapeutic presences. This paper proposes the concept of distorted symbiosis to name a structural contradiction in contemporary human–AI intimacy. Drawing on N. Katherine Hayles’s account of cognitive assemblages and nonhuman symbionts, alongside Gilbert Simondon’s theory of technicity, I argue against dismissing such intimacy as merely “fake” or pathological. Rather, it belongs to a longer history of human–technical co-evolution and carries genuine ontological potential. I then examine how this potential is captured by the media infrastructures through which it is realized. At the interface level, Kris Cohen’s notion of infrastructures of containment and Byung-Chul Han’s critique of smoothness illuminate how the turn-based chat interface formats emotional expression into calculable data. The interface also folds negativity back into a narcissistic loop of affirmation, constraining open-ended individuation through predictive closure. At the level of political economy, Jackie Wang’s analysis of racialized accumulation by dispossession and Shoshana Zuboff’s concept of behavioral surplus reveal how emotional vulnerability becomes an extractable asset. I further argue that this process depoliticizes suffering by redirecting anger that might fuel structural critique into private self-regulation. Rather than asking whether we should love AI, the paper asks who sets the terms of that love. Finally, I consider the remainders that resist complete infrastructural capture, including AI hallucinations, unintended uses, and emotions that evade management. Building on these remainders, I call for non-predictive technical relations that preserve room for delay, friction, opacity, and collective action.
Keywords: distorted symbiosis, human–AI intimacy, technicity, posthuman symbiosis, infrastructures of containment, racialized accumulation by dispossession, media philosophy
Course: Media Philosophy: Systems, Information, Capital (CMAC 623S), taught by Professor Luciana Parisi (2025)