I Built My First AI Chatbot at 15 and It Actually Works!!

Student builds AI chatbot at 15
Students Blog

I Built My First AI Chatbot at 15 and It Actually Works!!

Okay so I have been wanting to write this post for WEEKS because I cannot believe it actually worked. I built an AI chatbot. Me. A 15-year-old. With no prior experience in AI. And it answers questions about my school timetable in actual plain English. I am not going to lie, when it first responded correctly I literally screamed and my mum came running thinking something was wrong 😂

How It Started (AKA a Very Boring Sunday)

It was a Sunday afternoon and I was bored of revising for my geography test. I had been learning Python for about six months at this point — mostly following YouTube tutorials — and I thought, okay what if I could just ask my phone "when is my maths class?" instead of checking the timetable PDF every single time. Sounds simple right? It was not.

I spent the first weekend just understanding how chatbots actually work. I found out about something called intent recognition — basically the chatbot tries to figure out WHAT you are asking before it figures out HOW to answer. That blew my mind honestly. Like the bot does not understand words the way we do, it just matches patterns. Once I understood that, everything started to click.

What I Actually Used (Free Stuff Only!)

I used Python obviously. I used a library called nltk for the natural language processing bit (my older cousin explained what that meant). I stored my timetable in a simple Python dictionary — like a lookup table. And I used a free model from Hugging Face for the intent classification part. Everything was free. The whole thing runs on my laptop.

  • Python 3.11 (downloaded from python.org)
  • nltk library for text processing
  • A pre-trained small model from Hugging Face
  • VS Code as my editor (also free)
  • About 200 tabs of Stack Overflow
The biggest lesson: when something breaks, read the error message. Like actually read it. It usually tells you exactly what went wrong. I spent three hours on a bug that the error message was literally describing to me the whole time. 🤦

The Part That Took the Longest

The hardest part was not the code — it was figuring out all the different ways someone might ask the same question. "When is maths?" and "what time is my maths class?" and "do I have maths today?" are all the same question but they look completely different to a computer. I had to manually write out like 30 different versions of each question to train my intent recogniser. That took FOREVER. But it worked.

What I Want to Build Next

Now I want to make it answer questions about homework deadlines too — I could connect it to my Google Classroom. If anyone reading this is also learning Python and wants to build something similar, just start. Seriously. You do not need to know everything before you begin. I definitely did not. And if you get stuck, Stack Overflow and YouTube will save you. Every single time.

Arjun Sharma, Age 15
Arjun Sharma, Age 15

Arjun is a Year 10 student who discovered programming at 13 and has been obsessed ever since. He writes about his coding adventures, school projects, and everything he is figuring out along the way.