This study investigates why LLMs hallucinate when reasoning over linearized structured knowledge. It reveals that hallucinations stem from systematic internal dynamics, such as attention concentrating on shortcut cues and feed-forward layers failing to ground provided knowledge.
This paper argues that reducing disagreement in multi-agent systems is insufficient for value-laden tasks, proposing a knowledge-representation layer. This layer abstracts reasoning traces and agent decisions into symbolic disagreement states, distinguishing four types, with application in content moderation.
This paper analyzes a method for factorizing formal contexts into independent subcontexts using necessity operators from possibility theory. It studies the properties of these factorizations and extends the method to fuzzy contexts, enabling the computation of independent subcontexts in fuzzy data.
This paper investigates the size complexity and decidability of first-order progression, a method for updating knowledge bases in AI. It demonstrates that for specific action classes within the Situation Calculus framework, first-order progression grows polynomially and maintains decidability when using certain knowledge base fragments.
This article explains why general AI like ChatGPT fails at answering Islamic questions, as it is trained on unfiltered internet data and cannot cite verified sources like the Quran or Hadith. It generates statistically likely but unverifiable answers, which is problematic for a religion where the source of knowledge is as crucial as the knowledge itself.
A new automated pipeline, SkillX, improves the performance of autonomous language-model agents by extracting reusable, hierarchical behaviors from collective trajectories. This three-tiered knowledge base (strategic, functional, atomic skills) allows weaker models to efficiently retrieve experiences, overcoming limitations of traditional methods.
This practical guide explores ontology in computer science and AI, describing it as an essential framework for organizing knowledge. It enables machines to interpret relationships and make more accurate decisions, being crucial for semantic systems and next-generation AI applications. Leaders like Salesforce emphasize its importance for personalization and decision intelligence.
KARITA (Knowledge-driven Augmentation and Retrieval for Integrative Temporal Adaptation) is a system developed to address the challenges of temporal shifts in AI models, which are trained on historical data but deployed on future data. It integrates knowledge-driven augmentation and retrieval to capture diverse shifts and leverage insights for improved temporal adaptation across multiple domains.
This paper reveals the pervasive phenomenon of "tool overuse" in LLMs, where models unnecessarily use external tools. It identifies a "knowledge epistemic illusion" and proposes a direct preference optimization-based strategy that reduces tool usage by 82.8% while improving accuracy.