This repository develops a research governance framework for AI-mediated human systems.
- Fieldlight Human Research Utility — a participant-governed institutional and technical layer for time-bounded, compensated, and auditable human-AI research.
AI systems increasingly observe, infer, remember, adapt to, and intervene in human behavior. In many cases, these systems blur the line between product experimentation, behavioral research, personalization, surveillance, and human subjects research.
What should human subjects research mean when AI systems can observe, remember, infer, adapt, and intervene?
- What is meaningful consent in AI-mediated systems?
- What are the boundaries of observation?
- What rights does a participant have to their own data?
- What constitutes harm?
- Who decides?
- What does withdrawal mean in systems with memory?
- What is authorship when user behavior becomes part of the training signal?
- Research questions
- Working papers
- Consent patterns
- Risk assessment templates
- Participant rights checklists
- Memory withdrawal protocols
- Case studies
- Governance diagrams
The Human Research Utility proposal identifies an initial canonical artifact set:
- Participant Charter — baseline rights that govern all human-AI research.
- Research Participation Grant
- Progressive Consent Protocol
- Compensation And Benefit Standard
- Study Closeout Record
This repository is public for reading and citation, but it is not open-licensed by default. See Copyright And Use.