Deep learning is the ability for an artificial autonomous operator to rely entirely on an algorithm that teaches itself to operate after having watched a human do it. An AGI outfitted with deep learning technology, uses pattern recognition protocols in its operations.
The black box
According to Will Knight, it's not completely clear how an autonomous car makes its decisions. Information from the vehicle’s sensors goes directly into a huge network of artificial neurons that process data, then delivers the required commands to operate the steering wheel, the brakes, and other systems. Knight further explains that the system is so complicated that even the engineers who design it may struggle to isolate the reason for any single action. There is no obvious way to design such a system so that the AI can always explain why it does what it does.