-- Must Read Links --
Growing Up With Alexa. MIT Tech Review.
This story explores something not many people are talking about at the moment - what will the impact of all these A.I. devices be on our children? If you read the recent Atlantic article "Have Smartphones Destroyed a Generation," it's a valid concern whether A.I. is going to make the next generation better off, or worse.
How To Regulate Artificial Intelligence. New York Times.
A great article by Oren Etzioni, head of Allen A.I. Institute, that proposes 3 laws for A.I.s. They are: 1) An A.I. system must be subject to all the same laws as its human operator. 2) An A.I. must always disclose that it is not human. 3) An A.I. cannot keep confidential information without approval. Go read the whole thing.
AI Writes Yelp Reviews That Pass For The Real Thing. Engadget.
This is interesting because the ability to replicate human actions like this online will fundamentally undermine a lot of the trust we have in place because of them. In fact, A.I. made lead to an arms race around the concept of trust because of tools like this.
Transformer: A Novel Neural Network Architecture. Google Brain Blog.
Google, which is already leading in neural machine translation techniques, is back with some ideas on how to make it even better with a new architecture. The gist of it is that, using a mechanism called "attention," the neural net has more context around each word.
Documents Detail Deepmind's Plans for A.I. Research on NHS Data. Techcrunch.
This is interesting because it discusses an issue that we will see more frequently in the future - data privacy, government data that is highly personally identifiable, and A.I. research. The benefits are large, but so are the risks for misuse.