Experimental platform for AI value formation — seeking collaborators

[Freeman Dyson]

I’m building a modular system to model value change in agents through social and emotional exchange.

Applications:

• AI alignment experiments in moral development

• Empathy and cooperation modeling

• Social exchange and reciprocity research

• Long-term value drift studies

Core design:

• Interactions quantified via give /​ take /​ receive /​ allow

• Empathy and energy balance weight decisions

• Base (long-term) and current (short-term) states

• Decay and growth thresholds for evolving morality

• Outcomes emerge from accumulated experience

Status:

• Backend: JavaScript + SQLite

• Real-time UE5 integration

• Narrative and symbolic overlay

• Ready for insertion of research-driven behaviors and logging

Seeking:

• Simulation scientists

• Cognitive modelers

• AI alignment researchers

• UE5 developers

Why collaborate:

• Theory-agnostic

• Swap in new models without rewriting

• Test how agents learn morality from interaction

DM or comment if interested.