Abstract:

In this paper, we describe a teaching experience in the subjects of Logic and Concurrent Programming of the Degree in Computer Science Engineering, on the use of ChatGPT 3.5 for autonomous learning and critical reasoning by the students. In particular, ChatGPT was asked to generate multiple-choice questions on various topics of these subjects, finding that in most of them this generative AI engine made mistakes both regarding the structure and the correctness of the questions. Furthermore, when the generated questions were incorrect and ChatGPT was warned through the interactive prompt about the error committed, it was not generally able to correct the errors. In light of these results, it was proposed as a teaching activity for the students to discern, on the batch of questions, which ones were correct, which ones were badly structured, and which ones were incorrectly solved. In this way, the students became aware of the limitations of the tool, while applying with a critical attitude the knowledge acquired in the subjects. As a result of this activity, the students changed their perception of ChatGPT from an automatic resolution tool to a learning support assistant.

Keywords:

ChatGPT, teaching experience, Computer Science, critical reasoning.