Artificial Intelligence Assisted Failure Analysis: Automated Extraction, Classification, and Root-Cause Determination. A Study Assessing the Capability of Current AI Tools

The research presented here introduces a practical approach for using artificial intelligence (AI) to review large collections of past failure analysis reports to help failure analysts identify probable causes of failure more quickly and consistently. The method uses AI tools that can read text, interpret data and figures, and learn to analyze this information to determine the root cause. It also includes image‑analysis tools that can recognize features such as fatigue striations, fracture modes, corrosion, and loading/overload characteristics in fractographs.

Combining information from the written reports, images, and numerical data, the AI system was challenged to group similar failures together, highlight conditions that commonly appear before specific failure modes, and suggest possible root causes. The system was tested using real historical reports describing various fractures modes and root causes.

Early results show that the AI is able to correctly identify major failure categories and can point engineers toward the factors that had the strongest influence on its conclusions. These explanations help ensure transparency and allow human experts to verify the results. This work demonstrates how AI was used to support failure analysts by reducing review time, and uncovering trends that may be difficult to detect manually. The long‑term goal is not to replace expert judgment, but to provide engineers with a powerful tool that helps them learn from past investigations and make better-informed decisions in the future.