Where is AI in the curriculum?
In the second-year Bachaloria Programming curriculum AI appears twice: in unit 1 (how it works, its uses and its ethics) and in unit 7 (machine learning, neural networks and large language models). This explanation is ours, to help you get the idea; the reference is the ministry's textbook.
The idea in one sentence
With an ordinary program you write the rules. With AI you give it examples, and it works out the rules itself.
Example: you want a program that tells a cat photo from a dog photo. It is hard to write a rule for every ear shape. Instead you show it thousands of photos labelled "cat" and "dog", and it learns the difference.
In your life without you noticing
- The videos YouTube and TikTok show you.
- Unlocking your phone with your face.
- The map telling you the road is busy.
- In industry: cameras that spot faulty products in a factory, and systems that predict when a machine will break down.
Kinds of machine learning
- Supervised learning: examples with answers (a photo plus "cat"). The most common kind.
- Unsupervised learning: examples without answers; the program groups similar ones together.
- Reinforcement learning: the program tries, gets a reward or a penalty, and improves, the way you learn a game.
One important thing: data is split into training and testing. If the program memorises the training data and cannot handle anything new, that is overfitting, like a student who memorised past papers without understanding them.
Neural networks and deep learning
A neural network is layers of small "cells". Each cell takes numbers, multiplies them by weights, and passes the result to the next layer. Training means adjusting those weights little by little until the answers come out right.
Deep learning is a neural network with many layers. It is what let computers understand images and speech.
Large language models (like ChatGPT)
A large language model is trained on a huge amount of text to predict the next word. Doing that word after word produces whole passages. That is generative AI: it generates new text or images.
But it can be confidently wrong. It can invent a fact that sounds right and isn't, so check anything important against a source.
Ethics
- Bias: if the data is biased, the program will be too.
- Privacy: whose data was it trained on, and did they agree?
- Transparency and accountability: if it gets something wrong, who is responsible?
- Fakes: fake images and voices that look real.
- Work: jobs will change, and new skills will be needed.
A practice question (ours)
A program trained on cat and dog photos scores 100% on its training data and 60% on new photos. What is the problem called? Answer: overfitting.
From someone who works with it every day
Eng. Ahmed Tarek is an AI product manager and founded an AI-powered learning platform. He teaches both units on the subject page, and his first five videos are free.