International Journal of Science, Engineering and TechnologyAuthors: Research Scholar Ashish Mulraj Ved, Assistant Professor Ravindra Kumar
Abstract: Artificial intelligence (AI) is becoming easier to install in workplaces than it is to use well. A camera can identify a missing hard hat, and a predictive system can flag a risky shift, yet neither result explains who should respond, how quickly action must follow, or whether the control actually worked. This article examines the often-overlooked interval between an AI-generated signal and a competent preventive decision. Drawing on a structured, non-exhaustive critical synthesis of occupational health and safety (OHS), human factors, training and AI-governance literature, it sets out a six-dimension Predictive Safety Management Framework (PSMF). The framework links data governance, AI selection and validation, predictive monitoring, human-AI collaboration, training and competency, and ethical governance with continuous performance evaluation. Its contribution is to make competency a continuing operational requirement rather than a single training event at system launch. A role-based competency matrix and six decision gates show what evidence an organisation should seek before moving from a promising demonstration to supervised use. The paper also proposes measures that separate alert quality from response quality and from verified control outcomes. This is conceptual framework development, not a field evaluation: no employees were surveyed, no training intervention was tested, and no injury reduction is claimed. The central argument is practical: AI assists prevention only when workers retain the knowledge, authority, time and trust needed to act on its information.
