<p>The ChatGPT tool, the large language model developed by OpenAI, is having a great impact among users, experts and scholars for its capabilities of answering questions and retrieving solutions automatically. Despite the short time since its release, it has already been employed in several application domains. However, to the best of our knowledge, it has not been studied in the field of operation research (OR). In this paper, we use ChatGPT to define solution strategies for addressing several variants of the shortest path problem. The results obtained by executing the solution approaches returned by the tool are compared, in terms of correctness and efficiency, to reference codes. They indicate that the proper utilization of this tool could represent a good aid for domain experts. In particular, the outputs provided by ChatGPT could represent not only a good base for more complex implementations, but also they represent a way to facilitate some tasks in order to reduce times to do certain activities, which in any case must involve human control, adaptation and supervision.</p>

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

Chatgpt and operations research: evaluation on the shortest path problem

  • Martina Luzzi,
  • Francesca Guerriero,
  • Marco Maratea,
  • Gianluigi Greco,
  • Marco Garofalo

摘要

The ChatGPT tool, the large language model developed by OpenAI, is having a great impact among users, experts and scholars for its capabilities of answering questions and retrieving solutions automatically. Despite the short time since its release, it has already been employed in several application domains. However, to the best of our knowledge, it has not been studied in the field of operation research (OR). In this paper, we use ChatGPT to define solution strategies for addressing several variants of the shortest path problem. The results obtained by executing the solution approaches returned by the tool are compared, in terms of correctness and efficiency, to reference codes. They indicate that the proper utilization of this tool could represent a good aid for domain experts. In particular, the outputs provided by ChatGPT could represent not only a good base for more complex implementations, but also they represent a way to facilitate some tasks in order to reduce times to do certain activities, which in any case must involve human control, adaptation and supervision.