What Makes AI Smart?

What Makes AI Smart?

Have you ever wondered how… how exactly AI works?

Do someone from the backend is giving answers to our question? or it is computer generated output??

So welcome to the Aviral’s Code Blog!

Where I promise to answer all the questions you have in your mind after reading the lines above…

In the my second last blog, I have discussed how AI is different from humans. And I am sure if you had read it, you must be curious about the real working of AI, what makes a “Dumb” computer “Smarter”?

So… Let’s dive in 😉

First of all, who are you? Why are you so smart😎??

Yes yes it’s your quality… but if you are smart, then tell how? what makes “you” smart?

Before answering this for AI, let us find out what make humans that smart!

The answer is…. Brain… you got the point!

What is brain made up of?

Let me call one of my Intelligent 🧠 friend …

Heyyy, Aviral here!… I have a friend with me.. and wants to know… what our brain is made up of?? because I know we human are made of cells… but what makes my brain different from my hands?

Oh,, okk okk.. thanks ☺!

He said “Dear, Your brain is made up of specialized cells called neurons and glial cells, which allow it to process and transmit information. Unlike your hand, which is made mostly of muscle cells, bone, and connective tissue for movement and structure, the brain's neurons form complex networks to control thoughts, emotions, memory, and body functions. Think of your hand as the worker and your brain as the boss giving instructions!”

Wait.. he said your!!! then what about him?? did you noticed?

Oho…my fault, I should have told you its name before calling… that was Chat GPT 😂!

But but but…

Your question was right, what about “its” brain?

What is the brain of AI?

As you know our brain is primarily made up of neurons, similarly AI also have neurons, and many neurons together called “Neural Network”!

So what are neurons and what are they made up of?

No we are not going into biology again… let’s stick to Technology purely.

Basically we have some mathematical functions, written in form of code, and we say them neuron!

This mathematical functions takes input, process and gives output.. as simple as that.. But if we dive more deeper, there are many complicated formulas and algorithms working inside. But if I am going to talk about them here, it will take twice the length of this blog!

So let’s summarize before we move further, we have a set of algorithm which takes input and then work on it then gives output, and we say it “a neuron”.

Now you might think, what does it actually do with the input? what is the output? and what it take as input?

So answer these question, let us think of a Image recognition model, which tells whether its a cat or not…

So the input will be a pixel, and the neuron will try to figure out that how much this pixel contributes to a cat’s image on some specific parameter using some mathematical functions, now, the process is already started and each neuron is checking pixels according to its parameter and sending output to the further layers, and the final layer will be giving the output! this means, after analyzing all the pixels, the final output will be “Yes” or “No”, or to be more precise, 1 or 0!

So let me clear the picture more, A Neural network is made up of many layers, mainly divided into 3 parts, Input layer, Hidden Layers, and finally Output layer. The input and output layers works as the name suggest, and hidden layers may have 1 or more layers in it depending on the complexity of the model. And each layer is consist of multiple neurons! Now we have a neural network with some layers consist of many neurons, which can be mapped as,

Here each circle is a neuron, lines represent output and vertical groups of neurons are layers.

So in this blog we discussed about what makes a machine smarter.

If Curious about how these neural networks are created? Stay tuned for my next blog.

—spoiler alert: it involves a lot of coding fun!"

Till then keep exploring for fun!

- Aviral Srivastava