Great feature from Will Knight on AI’s language problem
. “If AI is to serve as a ubiquitous tool that people use to augment their own intelligence and trust to take over tasks in a seamless collaboration, language will be key.”
Cathy O'Neil: Trump in an object lesson in machine learning
. “Trump’s algorithm is to say semi-random things until his crowd roars its approval, then he iteratively modifies those statements, seeking more and more approval, until he maxes out and tries a new tack.” Makes a great point about the risks of our increasingly algorithmic future - systems designers need to think about what their systems end up optimising.
Vector-space maths formally uncovers gender bias
in text. Super application of word2vec helps us identify hidden (often uncomfortable) relationships between words as use in ordinary written language. The sub-headline of this article is amusing: “As neural networks tease apart the structure of language, they are finding a hidden gender bias that nobody knew was there”, as one Twitter user wrote the sub-head should read “As neural networks tease apart the structure of language, they are finding a hidden gender bias that EVERYONE IN THE HUMANITIES KNEW WAS THERE
.” (My emphasis.)
Huge reading list of papers supporting “Critical Algorithm Studies
”. Haven’t had a chance to do more than skim the list, but looks fascinating.
Elsewhere: Intel acquires Nervana Systems
which makes chips optimised for deep learning. More evidence that we are in the deployment phase of artificial intelligence. (Congrats to the several EV readers involved as investors, advisers or employees of Nervana Systems.)