Hallucination and Structural Reliability

Hallucination as a structural property of generative models: its causes, patterns and the limits of reliability, together with mitigation approaches such as grounding, citation and confidence calibration. Excludes building retrieval-augmented generation systems, which belongs to 8.3, measuring hallucination with metrics and benchmarks, which belongs to 9.1, and the societal harm of misinformation, which belongs to 12.5.

• 8.3 RAG, Tool Use and Memory
• 9.1 Measurement Frameworks and Reference Metrics
• 12.5 Democratic Space and Freedom of Expression

• 5.1 Foundation Models and General-Purpose AI
• 5.2 The Transformer and Beyond
• 5.3 Mamba, State-Space Models, MoE and Emerging Architectures
• 5.4 Pre-training and Scaling Laws
• 5.5 Fine-tuning, RLHF, DPO and ORPO
• 5.6 Context Protocols and Tool Use
• 5.7 Multi-Agent Orchestration
• 5.8 Open, Open-Weight and Closed Models
• 5.9 Natively Multimodal Models
• 5.11 Sovereign and Localized Models