DEV Community

Alen Kurian
Alen Kurian

Posted on

I Thought AI Actually Understood Language… I Was Wrong (Day 1)

Today I started my 90-day journey into Artificial Intelligence.

No hype, no shortcuts — just learning step by step and sharing what I understand.

What I Learned Today

I began with Large Language Models (LLMs).

Before today, I genuinely thought AI understands language like we do.

But the reality is very different.

LLMs don’t understand language.They predict the next word.

How It Actually Works

At its core, an LLM is trained on massive amounts of text data.

Given an input like:

"AI is changing the..."

The model predicts the most probable next word, such as:

"world"

It does this by learning patterns and probabilities from data — not by understanding meaning.

Every sentence generated is just a sequence of these predictions.

What Surprised Me

Even without true understanding, LLMs can:

Write essays

Answer questions

Generate code

This made me realize:

Intelligence (in this case) can emerge from simple rules applied at scale.

My Takeaway

What feels like “thinking” is actually:

Pattern recognition

Probability

Massive training data

And somehow, that’s enough to create powerful AI systems.

What’s Next?

Tomorrow I’ll explore:

What tokens are

How text gets converted into numbers

Day 1 complete. 89 days to go.

If you’re also learning AI, let’s connect and grow together.

Top comments (0)