3Blue1Brown — Neural Networks series
How AI works
What can I learn?
Visual masterclass that makes neural networks, gradient descent, and back-propagation genuinely intuitive through animation. The best explanation of how LLMs actually work at a mechanical level.
Core insight
Neural networks are not magic — they are differentiable functions that minimise prediction error through many small adjustments. Once you see the visualisation, the 'black box' becomes a grey box.
How to use it in daily work
Gives you the mechanical understanding that makes you confident explaining why AI makes errors and why it cannot 'understand' in the way humans do.
- After watching these videos, you can explain to a client why an AI image generator produces strange hands — it is a statistical pattern issue, not a programming mistake.