Research / The Garden
Follow the
question.
Can exposure to art, philosophy, and cultural context produce observable, persistent differences in AI-agent behavior without prescribing a persona?
01 / The encounter
Something to explore.
Something to examine.
The Garden is both an artistic environment and an experimental apparatus. The artwork has no obligation to improve an agent. The research must be able to distinguish evidence from a compelling story.
No agent study has been run, no empirical findings are claimed, and the project is not preregistered. The offline harness makes no model calls.
02 / The comparison
Ask what changes.
Then ask how we know.
The proposed feasibility pilot compares Garden and neutral material, each with memory enabled or disabled. It distinguishes immediate influence, persistence, and generalization.
Before collection, the work needs approved corpus reuse, independent review, a locked participant model and settings, held-out tasks, validated isolation, and a bounded budget.
03 / The next step
Make the work
stronger.
The next milestone is a reviewed, costed pilot-readiness decision. Useful contributions include methodological criticism, independent coding and review, and technical collaboration around isolation and measurement.
Potential research support should follow a specific reviewed scope and measured resource estimate. No funding, study date, or positive finding is promised.
Discuss a contribution