Humanities Approaches to Computation
Computational systems as objects of humanistic inquiry.
Overview
ARCH supports humanities research on computing systems, including artificial intelligence models, agent harnesses and loops, big data systems, cybernetic and automated systems, code, hardware infrastructures, and the institutions that build and deploy them.
Our work asks how humanities disciplines can contribute directly to the study of a world where computing is now ubiquitous. Philosophy brings tools for analyzing and clarifying characterizations of reasoning, agency, consciousness, autonomy, responsibility, value, ethics, alignment, safety, fairness, and bias. Linguistics and Languages contribute to the study of reference, meaning, inference, ambiguity, and model behavior. Rhetoric, communication, and literary studies bring expertise in persuasion, audience, narrative, genre, figuration, voice, style, and interpretation. These humanities approaches aid mechanistic interpretability, model evaluation, and the study of AI agents as systems shaped by prediction, feedback, memory, tool use, and interaction with constellations of human language, human image-making, and human movement. ARCH treats computing not only as a tool, but as an object of research in its own right.
Research directions
- Philosophical analysis of agency, intentionality, and personhood in language-model-based agents.
- Humanities-designed AI benchmarks for interpretive and moral-reasoning tasks.
- Humanistic contributions to mechanistic interpretability.
- Definitions of reasoning; assessments of the relationship between reasoning and perception.
- Meaning, reference, and language formation in large language models.
- The social formation of human–AI interaction.
- Cultural aspects of gesture in teleoperation and robotics.
- AI ethics and alignment as problems of value and interpretation.
- Global cultural dimensions of model and agent behavior.