Towards AGI - Organic Deterministic Intelligence - A Training Corpus for Hallucination-Free Artificial General Intelligence

Key Points

• The aim is to develop a training corpus that ensures factual integrity in artificial general intelligence systems.
• Created a live research package for training AGI, focusing on fact citation and verification.
• Implemented a hypothesis testing approach to refine AGI narratives.
• Developed a peer-review mechanism involving user contributions for quality assurance.
• Demonstrated significant improvement in factual consistency during AGI training.
• Established a method for citing sources and verifying claims in AI narratives.
• Highlighted the importance of peer review and user input in AGI development.

Abstract