Finding Health & Safety Buried Treasure with AI

Abstract

The challenge to glean understanding and insight from an array of historical safety-related reports and observations has existed since the dawn of the HSE discipline. While most organizations today use traditional methods to analyze past events and activities along structured elements (time, place, risk rating and so on), a vast amount of wisdom around hazard identification, root causes and risk control measures remains buried in textual descriptions and reports, and teachable moments become lessons lost. The hands and minds that developed these textual artifacts may be among the most seasoned in the organization, bringing years of experience to bear on the issues and opportunities involved. Such artifacts are then clearly buried treasure. Exploring and surfacing the insights contained in artifact repositories calls for new tools. Using these, a new type of H&S performance indicator could emerge: latent indicators, lying concealed within the written record, offering as much or more value as the leading and lagging indicators used today. This paper describes leveraging the power of artificial intelligence (AI) to absorb large amounts of safety-related textual information, find common themes and identify similar events, which are then analyzed for patterns in causes and controls. This solution, used in concert with traditional analytics, offers unprecedented power to comprehend and visualize collective safety knowledge from historical record. Transforming words to wisdom in this manner not only illuminates the past but also provides a basis for actioning improvements in operational excellence.