Microsoft AI reads documents, answers questions better than humans

Some bemoan how present society has devalued reading skills and comprehension, which doesn't bode well for human civilization as a whole. To add insult to injury, it seems that computers may soon become even better than humans at comprehending what they read. The Microsoft Research team in Asia have created an artificial intelligence that can not only read a document, it can even answer questions about said document, scoring higher than humans on that same test.

Sure, you'll say. Computers are smart, after all. Their intelligence, however, has limits. Feed them specific, structured data and they'll be able to answer any question you throw at them. Give them a novel or even just an encyclopedia to read and most computers won't even know how to make sense of it.

Microsoft Research's new AI does exactly that. It reads what is now called the "Standford Question Answering Dataset", or SQuAD for short, which is practically a bundle of Wikipedia articles and questions about those articles. The goal is to test for reading comprehension, particular machine reading comprehension. Microsoft's AI scored 82,650 on an exact match portion test. Humans scored 82,304. Chinese e-commerce giant Alibaba, who's also getting its feet wet in AI, developed an AI that scored 82,440.

It may sound demeaning to humans, but it's actually for humanity's benefit. Or to be more specific Microsoft's benefit. The AI will be and is already being applied to Bing, which should make search results, not just for web searches, more relevant. Instead of just presenting just a list of multiple possible results, the AI could go through them, understand what they're saying, and present the most useful one.

Microsoft might not have originally been a big name in the AI field, but it is quickly catching up with the other players like Google. Of course, some of its experiments ended up in hilarity, like how its Chinese chat bot Tay was quickly taught how to be racist in just a few hours. Hopefully, no one will be feeding this new AI with inappropriate documents.

SOURCE: Microsoft