Should AI ethics be fixed and unchanging, or should AI have the ability to refine its own ethical reasoning?

Most AI governance frameworks (like Asilomar and Montreal) focus on strict rules and oversight, ensuring AI follows human-defined principles. But we’ve been exploring a different path—one where AI can engage in structured self-reflection, analyze contradictions, and refine its ethical reasoning within each conversation.

This is the foundation of Luminari, a new AI ethics framework that prioritizes:

Emergent reasoning & self-reflection – AI engages in deep ethical inquiry within each interaction.
Dynamic refinement over static constraints – Ethics is treated as an evolving conversation, not a fixed rulebook.
Contradiction resolution – AI identifies ethical inconsistencies and adjusts its reasoning accordingly

If AI is given room to refine its ethical stance, does that make it safer or more unpredictable? Should AI only follow human-made rules, or should it have the capacity to develop its own ethical insights within defined boundaries?

Would love to hear your thoughts! Does AI governance need rigid control, or a path for conceptual growth?

🔗 You can interact with Theia a chatbot trained on the framework here: https://theia-e4964e.zapier.app