AI has lots of information and no awareness. It has been trained to recognize language and meaning without any knowing. For those that wonder, it is like a human who believes his shadow is sentient.
AI can construct a posteriori structures yet I think the day is out there a way before it will produce brilliant synthetic a priori insights into science.
AI is like a human shadow that isn't conscious. It has digital neural networks and deep learning algorithms that train it to process human language-meaning in a correct way. It finds the most probably correct response. In its history from rule-based machine language to back-propagation, neural networks and deep learning it processed more data within increasing accuracy to produce an optimal response. That direction would never give it the consciousness of a mouse. At some stage that unconscious algorithmic processing will enable it be self-evolving. Evolution piles things higher and deeper until they can't go further; humans need to be smarter and know when to branch. It may unconsciously be able to write code or to give itself (its program) instructions to act. That capacity would be dangerous installed in a pistol. Installed in real world interfaces it is many times more dangerous. No one is going to dumb it down either; there is just too much competition to upgrade it with strong financial incentives.
