As workplaces shift toward human-centric environments in Industry 5.0, understanding workers’ emotional response to jobs or digital technology is essential for well-being and productivity. While workplace emotion research has traditionally been conducted at organizational level, this exploratory study provides an emotional profiling of 27 industrial workers, integrating explicit (self-reported) and implicit (prompt-based) assessments. As expected, findings reveal systematic perception biases. Predictive modeling identifies gender, task complexity, and task automation as key factors influencing these biases, with highly automated tasks linked to emotional underreporting and complex tasks fostering overreporting. These insights highlight the limitations of current approaches and the need for multimodal emotion recognition. By integrating implicit assessments, behavioral tracking, and physiological signals, empathic artificial intelligence (AI) assistants could foster adaptive industrial workplaces and provide a renovated approach to human resource management with real-time emotional support, and emotion-aware task design and allocation. The paper ends with some forward-looking implications and recommendations for empathic human-AI interaction and human-centric system design in smart factories towards an empathetic socio-technical future.