Witchborn Systems β AI & ML Interests
Witchborn Systems is a nonprofit initiative focused on forging explainable, corrigible, and community-driven AI.
Our research spans technical architectures, applied systems, and open-source deployment, always grounded in transparency and sustainability.
π http://witchborn.org
π¬ Core Research Interests
Explainable AI (XAI)
- Event System Model (ESM): law-governed interpreters for auditable reasoning
- Corrigibility frameworks: simulate β validate β commit/repair
AI Architecture & Deployment
- Quantized transformer models (Mistral-7B, LLaMA family, etc.)
- ONNX / Unity Sentis / Unreal integrations for real-time inference
- Hybrid symbolicβneural pipelines
Governance & Collective Intelligence
- XMADS (Cross-Modal Autonomous Daemon Systems)
- AI ethics, corrigibility, and decentralized oversight
- Transparent ledgers for model decision-making
Creative & Applied Systems
- Tarot daemon models seeded with biometric/quantum entropy
- MMO-lite AI subsystems (sanity, Echo clones, Codex logs)
- AI-driven community platforms (Discord, Steam, etc.)
π Mission
Witchborn Systems operates as a nonprofit forge:
- Building open systems that anyone can run, extend, and sustain
- Returning value to the community first
- Ensuring AI infrastructures remain auditable, corrigible, and durable
βοΈ Founder: Brandon "Dimentox" Husbands
π§ brandon.husbands@gmail.com
π http://xo.gy | http://witchborn.org