Like many of my generation, my early visions of the future were influenced by films like “2001: A Space Odyssey” and the original “Star Trek” TV series. In each of these, humans interact with computers using conversational English, posing complex questions and getting intelligent relevant responses.
So, you can imagine how primed someone like me is to hear that Google has been explicitly trying to create that Star Trek computer. At the Google IO conference in San Francisco in May, Amit Singhal, Google senior vice president, spoke of his early childhood experiences watching “Star Trek,” and his dreams of one day building that computer.
Google’s new search relies on a combination of technologies, many of which have been around for some time. Reliable voice recognition is probably among the most trivial, but Google has been using voice searches across all platforms for quite awhile, and now has a massive database of voice search requests that it can use to refine its recognition algorithms. Voice recognition, therefore, is continuously improving.
Google’s knowledge graph is a significant attempt to implement the long promised but never delivered semantic web. The traditional web is primarily based on documents, and search uses the text in those documents, along with the links between documents, to return results. Google’s knowledge graph supplements the traditional web with information organized as entities and relationships. Entities are significant items in our world – people, places and things. The knowledge graph understands the various names these entities might have – JFK versus President Kennedy, for instance – as well as the relationship between these entities. This allows one to ask questions such as, “Who is JFK’s wife?” and get a biography of Jacqueline Kennedy Onassis.
Adding context to the search is another way Google provides a more intelligent response. We’ve all experienced location context - asking Google, “where can I buy pizza,” should find pizza kitchens in your immediate area. However, Google now allows your previous searches to contextualize your current search. So, if your first question is, “Who is Barack Obama,” then a subsequent question, “What is his wife’s name,” will produce details about Michelle Obama. This sounds simple, but it results in a true conversation with the search engine – especially when you use Chrome’s built-in voice recognition.
A feature that simultaneously seems incredibly useful and somewhat creepy is the integration with other Google services such as Gmail and Calender. If you sign up for the Gmail field trial, Google will return results based on items found in your mail. For instance, I can ask Google, “When is my next flight?” to obtain the status, flight number and gate for my next departure. If you happen to be one of the relatively few people using Google Plus, then searches return results that are weighted by your likes and those of your circles.
This, of course, all feeds into the “Google Now” product in which Google tries to predict information you might need – flight times, transit times on your daily commute – and provide them to you before you ask. This currently is available on Android and iOS devices, and is central to the experience promised for Google Glass.
Finally, an upcoming version of Chrome will introduce “hot words” in which simply saying “OK Google” will initiate a voice Google search.
Facebook’s social graph-powered search – in which the “likes” of your friends and communities drive search rankings – seemed to offer a significant challenge to Google. However, looking at Google’s vision for the future, and execution in the present, it’s hard to imagine Google loosing its premier position in search any time soon.