top of page

Revisit Forum Discussions

    My initial development plan was to use ChatGPT for programming assistance and turn to Godot’s forum when encountering problems that AI could not resolve. In practice, however, ChatGPT proved far more capable than I had anticipated. Compared with forum-based support, which involves waiting for responses and higher communication costs, AI provides immediate feedback and supports continuous follow-up questions. As a result, it became my primary problem-solving resource throughout the development process.

​

    After completing the project, I returned to the forum to explore how developers discussed Pong and whether alternative development approaches existed. Searching for the keyword “Pong” yielded only around fifty relevant posts, far fewer than I had expected. After sorting the results by the number of likes, I examined the ten highest-ranked posts. Of these, only two were directly related to the development of Pong, and they differed substantially in the depth of knowledge they provided. The more highly ranked post documented the entire development process, from software installation to project completion, while also explaining the logic behind key features and code implementation. By contrast, the other post merely shared the complete source code without explaining the development process or design rationale. While copying the code could help developers complete tasks quickly, understanding the relationship between the code and Godot’s nodes, as well as the underlying logic of its operation, still required additional self-directed learning. Most of the remaining posts consisted primarily of project showcases or brief discussions.

Two Posts about the Development of Pong

   At the same time, I searched the forum using the keyword “AI” to explore developers’ attitudes towards AI-assisted development. Among the most highly liked posts, one developer’s comment was particularly striking:

   This comment closely resonated with my own development experience: AI cannot replace a developer’s understanding of programming logic and development frameworks. Rather, the ability to use AI effectively has itself become a new form of competence. Developers must not only possess a basic understanding of programming concepts but also be able to break down problems, formulate effective prompts, and critically evaluate the quality of AI-generated outputs. These capabilities are not automatically provided by AI; instead, they are gradually developed through ongoing experimentation and accumulated experience.

​

   My experience with MuleRun further complicated this perception. With AI assistance, I was able to accomplish development tasks in a short period of time that would previously have been beyond my capabilities, and I could clearly perceive an improvement in my skills. At the same time however, I became aware that the very capabilities I had just acquired were rapidly losing their value.

bottom of page