The papers gathered here move across philosophy of technology and literary theory to interrogate the interpretive frameworks we bring to AI systems. Our papers claim AI models are themselves cultural artifacts, shaped by assumptions that they simultaneously inherit and distort.

1) Fabian Offert challenges the foundational premise that machine learning models are models of anything in a meaningful sense, proposing instead the concept of “epistemic alloy”—a hybrid form that partially represents the world and partially reproduces it wholesale, exposing the leaky abstractions embedded in scientific readings of AI.

2) N. Katherine Hayles brings her Integrated Cognitive Framework to reconsider the profound temporal asymmetry between human and LLM cognition: where human temporality is continuous, embodied, and historically saturated, LLM temporality is episodic and architecturally determined, with consequential implications for selfhood, cultural imagination, and the ethical responsibilities of human interlocutors.

3) Seth Perlow turns to aesthetic philosophy from Kant forward, to propose a parametric aesthetics capable of inferring facts about model behavior from the analysis of outputs, offering humanists a principled method for interpretation in the absence of direct access to high-dimensional parameter space.

4) Hannes Bajohr will serve as a respondent to the three talks.

4) Nina Beguš will preside over the panel.

Our session will begin with a short introduction of the topic and the panelist. Each of the three formal presentations will run for a maximum of 10 min, including the respondent's brief formal response to each (1-2 min), to allow plenty of time for the discussion. The presider will assure that the session runs on time and that we have both the discussion among the panelists as well as the audience Q&A.