Artificial Intelligence in EHS: 3 Myths and 3 Facts
Jean-Grégoire Manoukian
August 13, 2024
These were followed by integrated EHS software platforms that managed EHS processes within a single system and established single sources of truth. Then came mobile apps that allowed frontline workers to report incidents, near misses, and hazards in real time. Also, safety information was brought to the fingertips of those in the field.
Among the most recent innovations are data analytics for proactive safety assessments and actions. Insights generated by analytics help EHS teams identify risks and key trends more quickly and accurately.
The next big technological evolution in EHS is artificial intelligence (AI). AI is now being used for various EHS use cases, such as to deconstruct and derive important insights from regulatory documents, perform risk assessments, and provide predictive analytics.
But as is the case with all new technology, facts need to be separated from overexaggerated claims. AI in EHS is not immune from the hype and buzz that invariably accompany all groundbreaking innovation.
So, what can AI really do for EHS? This article explores three myths and corresponding facts.
Myth: AI Predicts When and Where the Next Incident Will Occur
Predicting with complete accuracy the day, location, and type of the next incident is like trying to accurately forecast the weather in your city a month from now. With so many uncertain variables – especially human factors – it is nearly impossible to prophesize with absolute certainty the next incident to occur. While some working areas may experience more frequent incidents and some types of incidents are most common, it’s impossible to know for certain in advance if and when someone might make a critical mistake, either due to negligence or organizational factors, that will result in an accident.Fact: AI Identifies High-Risk Areas Where an Incident Is Likely to Occur
AI may not be able to exactly pinpoint when and where the next incident will happen, but it can predict with reasonable likelihood areas where an incident is likely to occur. It’s like the weather. You may not know the exact weather forecast on a particular day a month from now, but you analyze current weather patterns and past weather history to have an idea of what may happen. When applied in EHS, AI can analyze incident data to identify potential areas, such as a plant, a specific process, and/or a work group, that may have a high risk for an incident. AI can also identify incident types that may reoccur (motor vehicle accident, slip/trip/fall, chemical release, etc.). Equipped with such knowledge, efforts can focus on high-risk areas through additional controls, more training, or other actions.Myth: You Only Need Large Data Sets to Use AI
Don’t make the mistake of believing an AI solution is only as good as the quantity of data available. Simply having more data doesn’t mean better results, insights, and suggestions generated by machine learning (ML) or natural language processing (NLP) models. Data quality is equally important and both quantity and quality are vital in the useful application of AI in EHS. Data is used to train AI models but, like other systems that turn data into insights, it’s important to remember that garbage in equals garbage out.Fact: Data Quality Is Also Key and AI Improves Data Quality
Before even considering AI for EHS, organizations need to ensure data is complete, consistent, accurate, and timely. You need to assess the effectiveness of your EHS data management systems and EHS data architecture. AI can be used to help improve data quality through ML and NLP by processing large data volumes and flagging inconsistent, duplicate, or incomplete records. These can then be edited or removed.Myth: AI Will Replace EHS Roles
AI will be used to improve EHS data, help organizations comply with regulations, and proactively identify safety risks. But ultimately, EHS leaders will retain responsibility and accountability for regulatory compliance and safety in their organization. The fact that AI may have recommended certain actions can’t be an excuse for noncompliance or workplace incidents. While AI can facilitate and accelerate processes and provide greater efficiencies, people will still need to verify, make final decisions, and take action.Fact: AI Will Evolve EHS Roles
Microsoft Excel offers a glimpse into how AI may impact EHS roles. Excel’s release in 1987 was a turning point for accounting and finance jobs. A Wall Street Journal article, We Survived Spreadsheets, and We’ll Survive AI, explains how, beginning in 1987, certain jobs, including bookkeepers, accounting clerks, and auditing clerks saw significant declines in demand. However, other jobs experienced great growth, including management analysts and financial managers. It’s reasonable to expect that AI’s impact will spur evolution in some EHS roles by simplifying complex, manual tasks, such as deconstructing legal documents (regulations, permits, etc.) to extract regulatory requirements and facilitate legal interpretation and applicability assessments.Expect a transition of certain EHS roles to more strategic areas, including proactive risk mitigation, specialized training, interpretation of insights, and ensuring AI is used effectively and ethically.
Related
Sponsored Content
About the Author
Jean-Grégoire Manoukian
Wolters Kluwer Enablon
Jean-Grégoire Manoukian serves as content thought leader for Wolters Kluwer Enablon, a leading provider of integrated software solutions that helps organizations protect worker safety, enhance sustainability, manage risks, stay compliant and identify
opportunities to elevate EHS, operational risk management, and ESG performance.