Google AI Responds To Questions Instead Of Simply Showing Search Results

Google has just developed an artificial intelligence that answers the questions of users directly instead of providing them with links to certain websites or pages that contain the information. For example, asking Google how fast a Peregrine falcon can fly will yield an answer along the lines of “389 km/h.” The process of actually providing this information is not a simple matter of just combing through stacks of data. Rather, it’s the result of a complex neural network that is meant to mimic how humans think.

The AI basically extracts relevant information from the stored data on Google’s servers, then breaks down the details to provide a more concise answer for users, Futurism reports. It’s basically like a human reading a book and then picking out specific information from said book when asked about it instead of showing the person the whole page.

Of course, this is easier said than done, since the deep neural network that Google is using makes use of compression algorithm to achieve the desired result. The company still succeed, since now, whenever users of Google’s Search app asks it for anything, it would simply give them the information directly.

As far as computer and artificial advancements go, this development is quite significant, Wired reports. Understanding human speech in terms of context and nuance has always been a challenge for computers. Even simply enabling the search app to answer questions pushed Google’s technical resources and capabilities to the limits says David Orr, the research product manager at the company.

“You need to use neural networks—or at least that is the only way we have found to do it,” Orr said. “We have to use all of the most advanced technology we have.”

It should be noted that more complex variations of questions will most likely be beyond what the AI is capable of right now. Even so, in the world of machine learning technology, the development so far is already considered substantial progress.