Artificial Intelligence as a Future Assistant for HSE Specialists

Today, artificial intelligence is one of the most popular areas of scientific research. Over the past 10 years, investments in this field have grown steadily, and in 2021, private investors made record contributions of approximately $125.4 billion. In 2022-2023, the volume of private investment in AI increased 18-fold compared to 2013. This type of technology is actively used in many sectors: from military affairs to the restaurant business.

According to research results, in 2022, the share of companies using artificial intelligence technologies more than doubled compared to 2017. Russian companies are actively following global trends in this area and are keeping pace. The volume of the Russian artificial intelligence market in 2022 amounted to 647 billion rubles, which is 17.3% more than the previous year. Many developments by Russian IT companies are used in the operations of domestic industrial enterprises. In 2022, about 17% of Russian enterprises were already using or implementing artificial intelligence, and most of them are large companies.

The global workplace safety market is projected to grow in the coming years. In 2022, it was valued at approximately $14.2 billion, and by 2027, it is expected to reach $26.7 billion. The market is set to grow at an average annual rate of 13.5% between 2022 and 2027.

Due to public demand for safe labor and the constant optimization of internal costs in most companies, IT solutions for automated monitoring of compliance with industrial safety and HSE requirements are becoming increasingly popular worldwide. Quality control systems for products and services based on video analytics are also highly relevant. Currently, the main directions of digital development in industry and manufacturing are the implementation of comprehensive solutions for detecting HSE violations based on computer vision and machine learning, as well as the development of methods to prevent potential problems.

When explaining the principles of currently relevant and available artificial intelligence technologies in the field of HSE, which include computer vision and machine learning, it is convenient to categorize them by their time of application. There are three such time zones relative to an incident and/or occurrence:

• Forecasting – using predictive analytics models to predict the likelihood of an accident based on the analysis of large volumes of historical data from similar cases.

For such a model to function successfully, it is necessary to collect data on past incidents, including workplace information (type of work, equipment, working conditions, etc.), worker details (age, experience, health issues, etc.), and the accidents themselves (type, severity, causes, etc.). By collecting this data over a long period, a model is created that analyzes the gathered information and identifies patterns and correlations between various factors. For example, the model may show that a certain type of work is the most dangerous, or that workers of a certain age with specific health issues are more prone to accidents.

• Real-time Analytics – rapid decision-making through the acquisition of up-to-date data.

To describe this technology, one can cite the example of AI that recognizes the absence of necessary personal protective equipment on workers and signals this through pre-arranged communication and notification channels. Such systems are currently capable of detecting and recording 95-98% of violations programmed into the AI algorithm. Furthermore, to better understand how the technology works, one can consider an example of a project implementing a system to inform about personnel presence in danger zones.

• Post-analytics and Root Cause Analysis – when evaluating past experience, it is possible and necessary to look for patterns of recurring causes and non-obvious correlations.

For instance, an enterprise may experience frequent accidents, such as those related to objects falling from heights. After collecting and analyzing incident data and training the model, it may be revealed that they all occur at the same time of day – during the lunch break, when workers leave for their break and do not monitor the situation at the workplace. The scenarios suggested by the model can be both obvious and non-obvious, allowing for a fresh perspective on old issues.

Artificial intelligence has enormous potential for application in industry and manufacturing. We are already seeing the results of using these technologies in large companies. Industry and manufacturing are only just beginning their journey into large-scale digitalization, and those who are the first to implement new technologies will win the competitive race. Despite the seemingly high cost of implementing solutions based on computer vision and machine learning, it is a long-term investment in business development. After all, these systems not only save money on repairing expensive equipment and minimize downtime but also allow for a transition to new methods of reducing worker injuries.

On one hand, it may seem that this field is too complex for HSE specialists to understand, and perhaps even redundant. However, the gradual widespread implementation of these progressive tools and methods, as well as the simplification of developing and preparing such models, will inevitably lead to an internal corporate demand for change. Therefore, in my opinion, modern HSE specialists should start improving their skills in these areas now to speak the same language as IT companies and their own IT departments when solving complex problems.