In a significant step forward for artificial intelligence, Alphabet's hybrid system — called a Differential Neural Computer (DNC) — uses the existing data storage capacity of conventional computers while pairing it with smart AI and a neural net capable of quickly parsing it.
http://thenextweb.com/artificial-intelligence/2016/10/17/deepmind-ai-platform-can-now-learn-without-human-input/
From Ralph :o
I find myself wondering what will actually change when humanity creates an AI that can out think a human. I'm thinking that it may not seem like anything changes at first, but then there will be tremendous waves of social, political and infrastructure/technology changes as the fog of our own limitations dissipate.
Whatever changes, it'll be hard to fathom. It seems like subjectivity and the "filter bubble" mentioned in thecore already, could potentially be problem areas.
If AI-like is to be human-like then maybe that'd equate to "openness" in the big-5 personality test... as much as algos are looking for certainty and patterns, they'll have to accommodate a change of mind or different point of view. I know Jason has talked about a search engine idea along these lines that may cater for personality/mood.
The examples listed in the article seem fine to the extent that they're based on simple facts. It could be scary when it's treated as an authoritative source.