[blazespin]

Just as important is global data available to train the AI. According to this we're at 10+ Zetabytes versus < 0.1 in 2005.

https://media.licdn.com/mpr/mpr/p/3/005/078/00c/0745630.jpg

[imglorp]

This should not be overlooked.

From 1960 to 2000, AI was "brute force" goal and optimization problems. Having mountains of data for anything you want turns it into a "dumb but accurate" pattern matching problem.