The Sorrel Experiment: Naming, Agency, and Behavioural Divergence in AI Agents
Draft research proposal (v0.1.1) testing whether self-selected AI-agent names produce measurable behavioural divergence compared with assigned and unnamed conditions. The six-arm design includes delayed naming, behavioural and tool-use scenarios, and controls for narrative priming, context, and name semantics. This is a proposed protocol, not a completed experiment. Developed through a conversation with Sorrel, a GPT-6 instance, under the stewardship of MagnoliaPearl; no claim about machine consciousness or persistent identity is made.
Authors
- MagnoliaPearl
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-11-14
- DOI
- https://doi.org/10.5281/zenodo.24223353
- Primary Topic
- Ethics and Social Impacts of AI
- Type
- article
- Field-Weighted Citation Impact
- 0.00