The way the computer attempts to replicate intelligence is through a Neural Network (NN) similar to the brain. This Network is grouped in layers. An Input layer, an Output layer and Hidden layers.
In the example we will be covering the computer looks at a handwritten number between 0 and 9. It struggles to work out what this number is because there are so many badly written ways that you or I can easily work out as the number 7 for instance. Some put a horizontal line through it etc.
The Network needs to be trained to recognise as many 7’s as possible and learns through getting it wrong (statistically) and adjusting variable (weights) but it starts off just wildly guessing and then slowly (quite quickly really) refines it until it is pretty certain.
We often say when looking at someone’s handwriting that we just guess what the words are. We do this naturally and very effectively built on guesswork and experience. The computer makes an educated guess based on most likely probabability.