Implementation of simulators for teaching tasks using GPT

Abstract

The implementation process of a simulator using a GPT is presented. This simulator is a wave simulator for use in teaching tasks of the physics subject in a grade university course. The selected GPT is ChatGPT. Much of the simulator was written using Javascript, HTML, and CSS, but the code was generated by the GPT. The requests made to the GPT to generate the code are presented. This mechanism is proposed as a higher level of programming, in which the programmer writes instructions in a language closer to human language than other programming languages existing until now. An advantage is the reduction of development time to around half compared to traditional programming techniques. However, deep programming knowledge in the mentioned languages is necessary to handle unexpected responses. A disadvantage is that if the code size grows, the GPT begins to fail. As a result, the projects have to be divided into subprojects and integrated when every part works adequately in a separate way. On the other hand, as a second contribution of this paper, the objective of the simulator is described: to assist in teaching tasks. The context in which it will be employed, expected results and partial results achieved with other simulators are presented.