2000 character limit reached
Beyond Prediction -- Structuring Epistemic Integrity in Artificial Reasoning Systems (2506.17331v1)
Published 19 Jun 2025 in cs.LO, cs.CL, and math.LO
Abstract: This paper develops a comprehensive framework for artificial intelligence systems that operate under strict epistemic constraints, moving beyond stochastic language prediction to support structured reasoning, propositional commitment, and contradiction detection. It formalises belief representation, metacognitive processes, and normative verification, integrating symbolic inference, knowledge graphs, and blockchain-based justification to ensure truth-preserving, auditably rational epistemic agents.