Development of an Artificial Intelligence Tool for Analysis of Process Safety Incidents and Extracting Learnings
This AIChE paper provides a direct process-safety example of RAG in operation. Air Products developed a tool that converts free-text incident narratives into structured classifications for incident cause and product loss. Its multi-RAG design retrieves both similar historical incidents and expert guidance notes. Those materials are inserted into a structured prompt, after which an LLM produces controlled labels that can be stored, trended, and analyzed. Validation against human-generated classifications achieved greater than 90% agreement, and the reported hose-burst example reduced an analysis that had required about 160 hours to less than an hour.
AI Engineering: Building Applications with Foundation Models
A widely used practitioner textbook covering how to build real-world applications on top of large language models. The RAG chapter is one of the most accessible technical explanations available: it walks through why retrieval is needed (models can’t know everything and can’t be retrained constantly), how documents are chunked and embedded into vectors, how a vector database finds semantically similar chunks, and how those chunks are inserted into a prompt so the model can “read before it answers.” Huyen also covers practical failure modes — irrelevant retrieval, context window limits, and the difference between RAG and fine-tuning — in plain engineering language, without heavy math.
Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks
This foundational paper introduced the term Retrieval-Augmented Generation and the basic idea that still underlies most RAG systems: combine a language model’s parametric memory with an external, searchable document index. A neural retriever selects relevant passages, and the generator uses those passages while composing an answer. The authors describe two formulations—RAG-Sequence and RAG-Token—and show improved performance and factual specificity on knowledge-intensive tasks.
Identification of Key Contributing Factors and Risk Propagation Paths in Safety Accidents at Chinese Chemical Enterprises
This paper demonstrates how text mining and network analysis of chemical industry accident investigation reports can identify systemic causative factors, reveal how they interact, and prioritize the risk controls most likely to prevent future process safety incidents.
Extraction and analysis of risk factors from Chinese Chemical Accident Reports
This paper demonstrates how natural language processing can convert unstructured chemical accident investigation reports into structured safety data, enabling more effective root cause analysis, trend identification, and continuous improvement in Process Safety Management.
A New Text-Mining–Bayesian Network Approach for Identifying Chemical Safety Risk Factors
This paper demonstrates how large-scale analysis of chemical accident investigation reports using text mining, association-rule mining, and Bayesian networks can uncover recurring safety risk patterns and support more effective incident investigation, organizational learning, and risk-based prioritization of corrective actions.
New Era of AI in Chemical Process Safety: Foundation Models
This paper explains how large language models combined with retrieval-augmented generation (RAG) can transform unstructured process safety records into structured, traceable incident investigation and root cause analysis outputs, enabling faster, more consistent organizational learning while maintaining the need for human oversight in safety-critical applications.
Domain-augmented large language models for automated root cause classification of offshore process incidents
This paper demonstrates how domain-adapted large language models can automatically classify root causes from offshore process safety incident reports, improving the speed, consistency, and auditability of incident investigations while supporting expert judgement and alignment with established root cause analysis frameworks.
Autonomous Fault Diagnosis and Root Cause Analysis for the Processing System Using One-Class SVM and NN Permutation Algorithm
This paper describes a self-learning AI framework that automatically detects process faults, identifies their root causes, and pinpoints the contributing process variables in real time, enabling earlier intervention and more effective incident investigations in complex chemical processing environments.
Information technology — Artificial intelligence — Management system (ISO/IEC 42001:2023)
ISO/IEC 42001:2023 is the first certifiable international standard for AI management systems, providing an auditable framework for governing, controlling, and continually improving the use of AI—including safety-critical applications within Process Safety Management—through defined policies, risk management, accountability, lifecycle controls, and continual improvement.
Artificial Intelligence — Guidance on Risk Management
ISO/IEC 23894:2023 provides guidance for identifying, assessing, and managing AI-specific risks throughout the AI lifecycle, helping organizations apply structured risk management practices to address issues such as model drift, data quality, transparency, and human oversight in safety-critical Process Safety Management applications.
OECD Framework for the Classification of AI Systems
The OECD AI Classification Framework provides a structured approach for categorizing AI systems according to their context, capabilities, and potential impacts, helping organizations determine the appropriate level of governance, validation, human oversight, and risk controls required for AI applications in Process Safety Management and other operational environments.
IOGP’s Principles for Responsible Use of AI with Guidelines for Implementation
This IOGP report provides industry-specific guidance for implementing and governing AI in high-hazard energy operations, with a strong focus on safety, human oversight, accountability, and risk-based deployment.
Regulation (EU) 2024/1689: Artificial Intelligence Act
The EU AI Act provides a detailed regulatory framework for governing AI systems, offering important guidance for organizations seeking to deploy AI responsibly in process safety, asset integrity, maintenance, and compliance-related activities.
Artificial Intelligence Risk Management Framework (AI RMF 1.0), NIST AI 100-1
The NIST AI Risk Management Framework (AI RMF 1.0) establishes a practical framework for identifying, assessing, and managing AI risks, offering process-industry organizations a defensible approach to deploying trustworthy AI in safety-critical environments.
Self-Consistency Improves Chain of Thought Reasoning in Language Models
Machine Learning and Deep Learning in Chemical Health and Safety: A Systematic Review of Techniques and Applications
Least-to-Most Prompting Enables Complex Reasoning in Large Language Models
AI and Process Safety Ethics
Fundamentals of Generative AI – Concepts and Terminology
This is a test of the excerpt in RSS.