Hello everyone,
I wanted to share a meaningful update on my solo research project.
QERRA-v2 Classical is now running with the complete SEMEV-12 ethical framework. All 12 vectors are fully active and contribute to a real weighted ethical score. Four of the most nuanced vectors (harm intent, cognitive manipulation, autonomy violation, and institutional trust) now use semantic detection via sentence-transformers, allowing the system to better understand subtle and context-dependent language.
The system also includes:
- Dynamic reasoning output
- Human-readable risk explanations
- Proper multi-vector activation
The public API is live and open for testing:
→ https://qerra-v2-api-classical-qerra-v2-api-classical.hf.space/docs
This is the classical (non-quantum) foundation of the project. It was built step-by-step as a solo researcher with the goal of creating a transparent, human-centered ethical decision system.
I would genuinely appreciate any feedback from the robotics and AI safety community — especially on how this kind of ethical reasoning layer could integrate with ROS2-based systems in the future.
Thank you for taking the time to read.
Best regards,
Marussa Metocharaki