Learn Python with ChatGPT: Write, Run and Fix Your First Programs

By Ravindra Bagale, trainer · LinkedIn · Last updated: 19 October 2026

Generative AI course · Chapter 34 · Estimated reading time: 13 minutes

1. Suppose Rahul wants to automate his shop bills

Suppose Rahul runs a small stationery shop. Every evening he adds up bills on a calculator and makes mistakes when he is tired. He has heard that "Python can do this", but he has never written a program.

He opens ChatGPT and types: "Write a Python program for my shop bills." He gets 60 lines of code, copies them, and something goes wrong. He does not know which line, or why.

Now suppose he types this instead: "I am learning Python from zero. Teach me to write a small bill calculator, one small step at a time. After each step, ask me to run it and tell you the output." Now he writes 10 lines he understands, runs them, and knows what each line does.

The tool is the same. The prompt decides whether you learn.

2. Where to run your Python

You need a place to run code. Two free choices:

Colab runs in your browser and is free to start with a Google account. It also has Gemini built in for eligible users: after an error, look for a button that offers to explain the error.

3. The tutor prompt

Save this prompt and start every learning chat with it:

You are my patient Python tutor. I am a complete beginner.
Rules:
1. Teach one small idea at a time, with a tiny example.
2. Do not give me the full solution unless I ask "show solution".
3. When I share an error, explain what it means in simple words,
   point to the exact line, and give me a hint first, not the fix.
4. After each step, ask me to run the code and tell you what happened.
5. Use simple English and short sentences.
Today I want to build: [your small project].

4. Worked example 1: a bill calculator

The prompt (after the tutor prompt):

Today I want to build: a bill calculator for my stationery shop.
Items have a name, quantity and price. Then there is a 5% festival discount.
Show me step 1 only.

Expected output (yours will be similar, not the same): AI shows a list of items, explains that a list holds many values, and asks you to print it. Over a few steps, you build something like this:

# Shop bill calculator: items, quantity, price, a 5% festival discount, total.
items = [("Notebook", 4, 40), ("Pen pack", 2, 55), ("Geometry box", 1, 120)]
DISCOUNT = 0.05
subtotal = 0
for name, qty, price in items:
    line = qty * price
    subtotal += line
    print(f"{name:<13} {qty} x {price:>4} = {line:>7.2f}")
discount = subtotal * DISCOUNT
print(f"{'Subtotal':<22} {subtotal:>8.2f}")
print(f"{'Discount 5%':<22} {-discount:>8.2f}")
print(f"{'Total to pay':<22} {subtotal - discount:>8.2f}")

Real output (we ran this on 19 October 2026):

Notebook      4 x   40 =  160.00
Pen pack      2 x   55 =  110.00
Geometry box  1 x  120 =  120.00
Subtotal                 390.00
Discount 5%              -19.50
Total to pay             370.50

Check it yourself. 160 + 110 + 120 = 390. 5% of 390 is 19.50. 390 minus 19.50 is 370.50. Always check the first result of any program by hand, even a tiny one.

Learning questions to ask AI next:

• "What does :<13 mean inside the f-string?"
• "Why did you write DISCOUNT in capital letters?"
• "How would I ask the user to type the items instead?"

5. Worked example 2: clean a messy CSV file

Suppose a teacher shares a student list exported from a form. It has extra spaces, mixed capital letters, a repeated student, a missing mark and the word "abc" where a number should be:

name,city,marks
 Priya Kulkarni ,pune,78
Amit Shinde,Pune,
priya kulkarni,PUNE,78
Rahul Patil , nashik ,abc
Neha Joshi,Nagpur,91

The prompt:

I have a file students.csv with columns name, city, marks. Problems: extra spaces,
mixed capital letters, one repeated student, empty marks, and text instead of marks.
Using pandas, show me code that cleans it step by step, with a comment on every line.
Do not delete bad rows silently: print them first so I can see what was removed.

Expected output (yours will be similar, not the same): code like the one below. We ran it on the file above:

import pandas as pd
df = pd.read_csv("students.csv")
print("Before:", len(df), "rows")
df["name"] = df["name"].str.strip().str.title()
df["city"] = df["city"].str.strip().str.title()
df["marks"] = pd.to_numeric(df["marks"], errors="coerce")   # "abc" becomes empty (NaN)
df = df.drop_duplicates(subset=["name", "city"])
bad = df[df["marks"].isna()]
print("Rows with missing or wrong marks:")
print(bad.to_string(index=False))
df = df.dropna(subset=["marks"])
df["marks"] = df["marks"].astype(int)
df.to_csv("students_clean.csv", index=False)
print("After:", len(df), "rows")
print(df.to_string(index=False))

Real output (19 October 2026):

Before: 5 rows
Rows with missing or wrong marks:
       name   city  marks
Amit Shinde   Pune    NaN
Rahul Patil Nashik    NaN
After: 2 rows
          name   city  marks
Priya Kulkarni   Pune     78
    Neha Joshi Nagpur     91

Look at what happened. Two students were removed because their marks were missing or wrong. That may be correct for a report, but those students still exist. The right next step is a human one: ask the teacher for Amit's and Rahul's marks. This is why our prompt said "print them first". AI-written code that quietly deletes data is the most dangerous kind.

6. Worked example 3: read a traceback and fix it

Suppose we write a small program to average some marks:

marks = input("Enter marks separated by commas: ").split(",")
total = sum(marks)
print("Average:", total / len(marks))

We type 78,91,64 and get this.

Real output (19 October 2026; we shortened the folder path):

Traceback (most recent call last):
  File "avg_bug.py", line 2, in <module>
    total = sum(marks)
TypeError: unsupported operand type(s) for +: 'int' and 'str'

How to read a traceback, from the bottom up:

1. The last line is the error type and message: TypeError, Python cannot add a number and text (str).
2. The line above shows the exact code: total = sum(marks).
3. "line 2" tells you where it is in your file.

The prompt (paste the whole traceback, not only the last line):

I got this error. Explain it simply, point to the line, and give me a hint, not the fix.
[paste the full traceback and your code here]

Expected output (yours will be similar, not the same): "input() always gives you text. After split, marks is a list of text pieces like '78', not numbers. sum cannot add text. Hint: turn each piece into a whole number before adding."

Now try the fix yourself:

marks = input("Enter marks separated by commas: ").split(",")
numbers = [int(m) for m in marks]   # turn each text piece into a number
total = sum(numbers)
print("Average:", round(total / len(numbers), 2))

Real output with 78,91,64 (19 October 2026):

Average: 77.67

Then test a bad input on purpose. We typed 78, 91, sixty:

    numbers = [int(m) for m in marks]   # turn each text piece into a number
               ~~~^^^
ValueError: invalid literal for int() with base 10: ' sixty'

A new error, and a new lesson: real users type words and spaces. Ask your tutor: "How do I show a friendly message instead of crashing when someone types a word?" (The answer involves try and except, a good next topic.)

7. The 4-step habit for every error

1. Read the last line of the traceback first, and say the error type out loud.
2. Find your line in the traceback and look at it before asking anyone.
3. Ask AI for a hint, pasting the full traceback and the code.
4. Fix it yourself, run it again, and then ask "how could I have avoided this?"

8. What AI gets wrong when it teaches Python

• Too much at once: a 60-line answer for a 10-line problem. Ask for "the shortest version a beginner can understand".
• Packages you have not installed: check every import line; install with pip install name if needed.
• Old or invented functions: if Python says a function does not exist, tell AI the exact error and your Python version (python --version).
• Confident but wrong maths: always check the first result by hand, like we did with the bill.
• Code that hides problems: like deleting bad rows without telling you. Ask it to print what it removes.

9. Beginner mistakes and quick fixes

10. Try at home

Recap

Write one 10-line program this week, break it on purpose, and fix it with a hint instead of a solution. Every error you understand is a lesson you keep.

Previous chapter: Vibe Coding for Beginners: Build a Website with AI (No Coding Needed)

Next chapter: AI Coding Tools Compared: GitHub Copilot, Cursor and Claude Code

Ravindra Bagale, trainer: linkedin.com/in/ravindra-bagale. The shop, students and marks on this page are made up for learning. Google Colab details were checked on Google's Colab pages on 19 October 2026. The Python output was produced on 19 October 2026 with Python 3.13 and pandas 3.0. AI output changes every time, so your output will be similar, not the same.