AI Fundamentals

Concept 1 of 12

Exit

Gradient Descent: The Core of AI Learning

How AI finds optimal solutions

Metrics

Steps:0
Distance:—
Loss:—
Key Insight

The orange arrow shows the gradient (slope). We move opposite to it. Watch what happens with different learning rates!

What is Gradient Descent?

Imagine you're in a foggy mountain valley and want to reach the lowest point. You can't see far, but you can feel which direction is downhill. Gradient descent is exactly this - taking steps in the direction of steepest descent.

1.

Calculate the gradient (slope) at current position

2.

Move in the opposite direction (downhill)

3.

Repeat until you reach the bottom (minimum)

Why Learning Rate Matters

Optimal

Reaches minimum efficiently in few steps

Too Slow

Takes many steps, wastes computation

Too Fast

Overshoots and diverges, never converges

Adaptive

Decreases over time for stable convergence