(in German)

Chatbots based on large language models (LLMs), such as ChatGPT, represent a specific form of artificial intelligence (AI). These language models are trained on large volumes of textual data to generate natural language. In the field of occupational safety and health, such chatbots can also find application - for example, by providing low-threshold access to information. They can serve as tools used both by employees and by those responsible for occupational safety.

In a collaborative project between Austrian Workers' Compensation Board (AUVA) and the German Federal Institute for Occupational Safety and Health (BAuA), usage scenarios, risks, and recommendations related to the use of publicly available chatbots in the context of occupational safety and health are being examined. Based on the literature and a user-centered analysis, exemplary use cases in the field of occupational health and safety are developed and evaluated by experts. The results serve as a basis for deriving recommendations that help users apply and interpret chatbot-generated answers appropriately.

The findings suggest that chatbots have the potential to support access to information in the field of occupational safety and health. However, due to the risk of incomplete or inaccurate information, verification and/or consultation with subject matter experts remains essential.

Practical Relevance: The results of the project provide practice-oriented recommendations for the use of generative AI chatbots in occupational safety and health. The developed guidelines and orientation frameworks aim to support workplace practitioners in deciding whether and to what extent chatbots may be suitable for their needs, and to facilitate an initial engagement with the topic.

The user stories illustrate the wide range of possible applications for chatbots - for example, in question-and-answer formats or the creation of informational materials. In addition, the developed model for classifying use cases helps identify appropriate areas for deploying AI chatbots and promotes targeted use. The recommendations for safe handling and interpretation of chatbot responses contribute to preventing misunderstandings and misinterpretations.

Please download the article "Chatbots for Occupational Health and Safety: Use Cases, Risks, and Recommendations" (in German only).