I was thinking about it a bit more and few questions came to me -- according to IBM, there is no strong AI as of yet. Considering that weak/narrow AI cannot handle what you are after in a reliable way, how long do you think will it take to get to (at least) general purpose AI (superintelligence being a pipedream for now)? Do we have so much time? And also, considering that AI, in essence, learns from human (directly or indirectly -- even the meta learning rules are human input) as well as the law of shit in--shit out, how do you assure that AGI is not just a computationally more effective but equally biased / ignorant mirror of human decision making? I really cannot wrap my head around the dilemma how a broken system could somehow produce a subsystem which would not inherit its parent's flaws.