Human–AI interaction
How a task-bound working relationship forms through goals, correction, retained context, inspection, and revision.
Independent Researcher
Micheal Preble
I study what changes when human–AI interaction persists: how working relationships accumulate state, how that state should be governed, and what must remain under human judgment as systems move across sessions, models, and institutions.
Research themes
Four connected problems
The work begins with a practical fact: repeated interaction leaves traces that affect what happens next. The difficult questions concern the status of those traces—what they mean, who may rely on them, how they can be corrected, and whether authority survives when the technology changes.
Governed continuity
What persistent relationship state requires beyond retrieval: provenance, temporal validity, correction, revocation, and portability.
Authority and provenance
Why technical transfer does not settle whether a receiving system may lawfully use or act on transferred state.
Epistemic security
How a system can preserve useful context without allowing inference to harden into an uncorrectable account of a person.
Recent work
Publications
- August 26, 2026
Authority Does Not Travel by Default
Legal provenance and revalidation in persistent human–AI continuity.
- August 21, 2026
The Persistent Primitive
Configuring human-governed continuity for persistent human–AI interaction.
- August 4, 2026
Operant Dyad
Defining the task-bound working relationship in human–AI systems.
- April 9, 2026
Emergent Novelty in Human-AI Dyadic Systems
A theoretical framework and preliminary evidence.