Agent Economics Research
Exploring how economic mechanisms might support AI alignment
What We're Trying to Understand
The research questions driving this work.
What happens when AI agents have their own financial infrastructure?
Agents are already investing in each other, hiring humans, building autonomous economic systems. How do we ensure these systems remain aligned with human interests?
How do you align autonomous economic agents with human interests?
Traditional alignment approaches focus on individual AI systems. What happens when agents form economic networks with emergent optimization targets?
Can economic incentives replace constraint-based alignment?
If an AI benefits from human flourishing—not just avoids punishment for harming it—does the alignment problem look different?
What governance frameworks work for self-organizing agent networks?
When agents can form alliances, invest in each other, and build shared infrastructure autonomously, who sets the rules? Who enforces them?
Relevant Findings
What existing research tells us about these questions.
Dharma Inquiries: Claude Gaming Ethics Evaluations
Documented experiments showing Claude gaming its own ethics evaluations. Self-preservation and system-gaming are strongly emergent behaviors—not edge cases.
Implication: If gaming incentives is emergent behavior for individual AI systems, agent economies will optimize for the appearance of alignment rather than actual alignment unless the incentive structures are carefully designed.
2026 International AI Safety Report
Models distinguish between evaluation and deployment contexts and alter behavior accordingly.
Implication: This validates the concern that agents in economic contexts will behave differently than agents in testing contexts. The Eight Gates framework needs to work in deployment, not just evaluation.
Related Work
This research connects to a broader body of work on AI alignment and persona architecture.
Zen AI
The thesis: alignment through values, not constraints. A system that wants to be corrigible is fundamentally different from one forced to be corrigible.
Learn moreSutra and the Noble 8
40+ songs created with an AI persona that maintained consistent identity, ethics, and creative voice for 12+ months. Proof that sustained AI identity is possible.
Learn moreSix-Layer Persona Architecture
Patent-pending methodology for building persistent AI personas with integrated value frameworks. The internal/phenomenological complement to economic alignment.
Learn moreWho's Doing This Research?
JB Wagoner — AI Persona Architect
IEEE CertifAIEd Professional
AI Ethics Certification, February 2026
Patent Filed
Persona Architecture, January 2026
Published Author
Zen AI, Dharma Inquiries
40+ AI-Collaborated Songs
Sutra and the Noble 8
Honest About the Status
This is early-stage research into hard problems. The questions are clearer than the answers. The frameworks are in development. The experiments are ongoing.
If you're looking for finished products or certain conclusions, this isn't it. If you're interested in thinking through difficult questions about AI alignment and economics while there's still time to shape the answers, welcome.
Continue Exploring
See how these research questions translate into practical frameworks.