🎶Meanwhile, some music for you to enjoy as you read. This piece is part of Vivaldi’s The Four Seasons, particularly the first of the 4 violin concertos, called Spring.
In our data-driven world today, quick and easy access to data for analysis is essential. However, challenges arise when data is not available promptly. In this article, I explore the use of AI models like ChatGPT, Bing Chat, and Bard to develop Python programs that tackle this issue. The goal is to create a segment of a Python program that checks for missing data files and enables manual report execution.
Background
Let's assume, we have monthly reports that generate data files for further processing. These reports are scheduled to run automatically. However, if the expected data files are absent from the designated folder, we need manual intervention to run reports to get the data.
Here is the prompt I started with.
Trying the AI Models
First, I went to ChatGPT
It identified a missing '' in a variable for a folder name.It effectively handled the file creation date requirement.
However, it did not understand the humor in the last part.
Next to Bing Chat
It also caught the missing '' in a variable for a folder name.However, it disregarded the creation date requirement entirely.
It show that it understood the desired functionality in the last part.
Then to Bard
It took the variable for the folder name with missing '' so that's not going to work.It didn't seem to understand what I was thinking for the last part.Additionally, it overlooked the file creation date requirement.
Program Comparison and Recommendations
So, next, I sort of go into a tree of thought mode by using three different "programmers" rather than as one programmer to come up with 3 solutions. Next, I decided to give the "programmers", the 3 programs and ask them to review and tell me the pros and cons of each and which of them is better. Unfortunately, I cannot run the prompt with the 3 programs to Bing Chat because of the 2,000 characters limitations.
I got through to ChatGPT using a new chat so it will not have any memory of the program it created before.
It thinks program 1 is the best. That's the one by ChatGPT.
I asked it to create a better program and it created one that sort of follow the suggestion of improvement it gave.
It defined a function and called the function which is nice.
Finally, it gave an explanation of the code.
Next, to Bard to review the 3 programs and also using a new chat so it will not have any memory of the program it created before.
It thinks the Bing Chat's program 2 is the better one.
However, when asked to rewrite a better program, it went totally off and got lost.
Conclusion
In conclusion, I was able to use AI models like ChatGPT, Bing Chat, and Bard to create part of a Python program that can be used to check for missing data files and prompt user to run reports manually. The programs created were not perfect, but it is a good starting point.
I believe that the collaboration between AI and human programmers is essential for the development of data processing tools. AI models can generate code, but they cannot replace human programmers. Prompt engineering to AI models to create program is not a silver bullet. We saw different AI models have their own strengths and weaknesses. Therefore, there is still a need for collaboration between AI and human programmers to achieve optimal results.
