My business is an independent videogame development company. Currently it is just me and ChatGPT, but the AI-powered tools provide automated code suggestions that can significantly speed up my programming process. These tools leverage large databases of code to suggest snippets that fit the context of the current project. This is particularly beneficial for beginner programmers like myself, who can learn from the suggestions and see best practices in action.
I have been working with ChatGPT for almost a year now developing code together, and I believe that I am understanding the language and verbage needed to squeeze meaningful code and help from the program. The AI allows users to describe functionality in natural language, which it can then translate into code. This can drastically reduce the entry barrier for individuals new to programming, allowing them to focus on the logic of what they want to achieve rather than the syntax of how to express it. Debugging is an interesting area when it comes to AI generated code, as AI is less prone to mistakes at what it thinks you want done, the largest problem I’ve found with AI code is that it may change variable or code names without warning. The other big issue is trying to explain what it is you want from the AI and iterating through what it thinks you want. For example saying you want an image on top of another has multiple meanings, do we want them one above the other? Do we want them overlapping? Finding where the disconnect is in the human prompts is the primary issue that is repeated in my work.