I think this path has flaws. First there is no idea why and how evolution developed consciousness inside our brains, and it's quite hard to simulate a black box

I think this path has flaws. First there is no idea why and how evolution developed consciousness inside our brains, and it's quite hard to simulate a black box only knowing the output. Secondly, once this problem is solved teaching the machine information processing would be similar to teaching someone else which fails because it seems impossible to get all rationals (moving target) properly communicated. Finally really _new_ news are rarely connected to existing data. To distinguish the interesting and surprising messages from just crap needs a different concept of how brains actually work.