Machine learning (ML) algorithms which are the core of AI are inspired by how we learn.
They are very different from the conventional code programmers compile, which consist of a well – defined list of instructions, which tell the computer what to do and when to do it.
For example, if you wanted to teach your dog to retrieve a ball, you wouldn’t produce a flowchart or a list of instructions for him to follow.
For one thing, you won’t be able to do so, because the way different balls bounce is unpredictable.
In any case, he wouldn’t be able to understand them!
What you do is throw the ball, and reward him when he retrieves it.
We reinforce good behaviour and ignore bad behaviour and give the dog enough practice until he finally figures out how to do it for himself.
And remember that learning how to retrieve a ball will not help him to learn how to shake hands – the training is very task – specific.
Similarly, in AI, you give the machine data, a goal and feedback when it’s on the right track-and leave it to work out the best way of achieving the desired end result .