“Exploring a Modular Ethics-Based Framework for Context-Aware AI: The AEPF Project (Seeking Feedback)”

post by ethoshift · 2025-04-21T08:40:22.400Z · LW · GW · 0 comments

Contents

No comments

I’ve been developing a non-commercial open-source framework called the Adaptive Ethical Prism Framework (AEPF). It attempts to guide AI systems using five interrelated ethical “prisms”, reporting into a central decision module (Ethical Governor) that adapts to context.

I’m not a professional coder—this is a passion project, and while the structure is in place (built in Python with a basic Streamlit UI), I know it needs a lot of refinement. The aim is to explore whether a modular, layered ethical structure can support better alignment by factoring in a range of contextual variables and perspectives (human, ecological, societal, etc.).

I'm sharing this in the hope of getting feedback—technical, philosophical, or otherwise—from others interested in AI alignment. I’d especially welcome thoughts on whether this prism-based modular structure has any potential in supporting safe and beneficial AI behaviour. \n\nGitHub: https://github.com/EthoShift/AEPF_Mk3
Website: https://www.etho-shift.com

Thanks in advance for any thoughts, critiques, or direction you might offer.

0 comments

Comments sorted by top scores.