Generolo Jono Žemaičio Lietuvos karo akademija

This paper deals with evaluation of chatbots abilities to solve tasks in operational research focused on assignment problems. Three chatbots – MS Copilot, Gemini, ChatGPT – were used for the research. The evaluation is based on the results of two model assignment problems. Selected characteristics of the task solving process were monitored. The characteristics are clearly summarized in tables. It allows us to determine the order of chatbots in terms of quality. Research has shown that using chatbots to solve assignment problems is possible. Gemini had the best results, followed by MS Copilot. The authors point out the need to give the chatbot very good instructions, force the chatbot to verify results, and use the right tools (independent zeros, cover lines, etc.).