Unit III · AI Tools for Coding

13. Debugging with a second AI tool

A bug is a mistake in a program. Debugging is finding and fixing it. In the lab, you will make a program with one AI tool, put bugs in it, and ask a different AI tool to fix it. Why a different tool? Because a second opinion catches more mistakes, just like two doctors.

Three kinds of bugs

Kind	What happens	Example
Syntax error	The code does not compile or run at all	Missing semicolon `;` in C, wrong indentation in Python
Runtime error	The program crashes while running	Dividing by zero, reading outside an array
Logic error	The program runs but gives a wrong answer	Using `<` instead of `<=` in a loop

Logic errors are the hardest. The computer does not complain. Only testing finds them.

How to ask AI to debug

Give the AI three things: the code, the error message (or the wrong output), and what you expected.

prompt

This C program should print the sum of numbers from 1 to N. For N = 5, it prints 10, but I expect 15. Find the bug, explain it in one line, and give the corrected code.

#include <stdio.h>
int main() {
    int n, sum = 0;
    scanf("%d", &n);
    for (int i = 1; i < n; i++) {
        sum = sum + i;
    }
    printf("%d\n", sum);
    return 0;
}

The AI will say: the loop should be i <= n. That is a logic error.

For a compile error, paste the exact error text:

prompt

I get this error when I compile: "error: expected ';' before 'return'". Here is the code: ...

The lab process (Experiment 10)

Generate. Ask tool A (say ChatGPT) for a program. For example, a Python program to find the second largest number in a list.
Run it. Make sure it works.
Break it. Put in 2 or 3 bugs on purpose. One syntax error, one logic error. Write down what you changed.
Debug. Give the broken code to tool B (say Gemini). Do not tell it where the bugs are. Ask it to find and fix them.
Compare. Did tool B find all your bugs? Did it find something extra? Did it change anything it should not have?
Run again. Check that the fixed program works.

Good debugging prompts

"Find all the bugs in this code. List each one with the line number."
"Why does this give a wrong answer for input 0?"
"Add print statements to help me see what the variables are doing."
"Explain this error message in simple words."
"Is there any input that will crash this program?"

Be careful

AI sometimes "fixes" code by rewriting everything. Ask it to change as little as possible.
AI may introduce a new bug while fixing the old one. Always test again.
If the AI says the code is correct but your output is wrong, check your input and your expected answer. Sometimes the bug is in your head, not the code.

Try it

Do the six-step lab process with a program of your choice. Use two different tools, for example ChatGPT and Gemini, or Gemini and Claude.
Paste a C program with a missing ; into OnlineGDB. Read the compiler error. Then give the same error to an AI. Compare what each one says.
Ask an AI: "Give me a Python program with 3 hidden bugs for practice. Do not tell me where they are." Then find them yourself. Then ask a second AI to check.

New words

Bug — a mistake in a program.
Debug — to find and fix bugs.
Syntax — the grammar rules of a programming language.
Runtime — the time when a program is actually running.
Logic — the reasoning or steps of a program.
Crash — when a program stops suddenly because of an error.
Error message — the text a compiler or program prints when something is wrong.
Second opinion — asking another person (or tool) to check the same thing.