Complexity Simplified #3: How Does AI Think?
- Amir Bder
- Jan 22
- 2 min read
Today, we’re looking under the hood at the engine that powers almost everything in the AI world: the Neural Network (fancy).
If you listen to tech CEOs, they make it sound like teaching quantum physics to a baby, but in reality, it’s just a very organized game of Hot or Cold.
The Fancy Words
A Neural Network is a computational model inspired by the human brain, consisting of layers of interconnected "neurons" (nodes) that process data by assigning weights and biases to inputs to recognize complex patterns.
Translation: It’s a giant filter that learns what you like.
Example: The TikTok For You Page
Think about the first time you opened TikTok or Instagram Reels.
The app knew nothing about you. It showed you a random mix: a cooking video, a cat dancing, a guy fixing a car, and a prank video.
The Input: You watched the cooking video twice but swiped away instantly on the car repair.
The Layers: The "Neural Network" inside the app noticed that. It didn't just think "Food." It looked deeper. It saw "Pasta," "Italian music," and "Short 15-second clips."
The Adjustment: Next time, it showed you more pasta. You liked that too? It turned up the "Weight" on Italian food content. You hated the prank video? It turned down the "Weight" on pranks (but whose watching videos of pasta...)
An AI works exactly like this. When you show an AI a picture of a Golden Retriever, it doesn't "know" what a dog is. It passes the image through thousands of layers (like the TikTok algorithm).
Layer 1 looks for edges.
Layer 2 looks for fur textures.
Layer 3 looks for "floppy ear shapes."
If the "Output" at the end says "Dog," the network gets a gold star. If it says "Toaster," the network adjusts all those filters (just like TikTok stops showing you car videos if you swipe fast) until it gets it right next time.
Why This Matters in Your Daily Life
Understanding Neural Networks removes the "spookiness" of AI. It’s not a conscious spirit; it’s a pattern-matching machine.
When Netflix suggests a movie: That's a neural network matching your "viewing pattern" to others.
When your iPhone unlocks with your face: That's a neural network recognizing the "pattern" of your features.
When ChatGPT answers a question: It’s just predicting the "pattern" of words that usually follows your question.
(do you get the "pattern" here... sorry that wasn't really funny I know)
The Bottom Line
AI doesn't "understand" the world the way you do. It just watches billions of "TikToks" of human data until it learns exactly what we want to see and hear.
Comments