By Adam Dickter / CIO Today. Updated May 22, 2013.
If you like chatting with Siri, sending voice texts while driving or telling your Xbox when to pause or rewind a DVD, you're going to enjoy the upgrade to Google's Chrome browser. The new system, which was previewed at Google's I/O developer's conference on May 15 and went live Wednesday, allows users to search for topics conversationally, by asking questions rather than stating keywords, much like Apple's Siri personal assistant for the iPhone.
Google also announced a stable update for its Android mobile operating system which allows entering topics in the omnibox, or address bar, instead of only in the search window on Google. The Android update also allows the toolbar to disappear while you scan results.
In addition, Google is giving Apple's Siri a run for her money by introducing voice search for the iOS version of Chrome, available "soon," that will allow voice searches on that browser for iPhones and iPads. (Siri will be available for iPads with the iOS 6 update.)
The Future Is Now
The update for the Chrome browser is another step toward the kind of interaction with machines seen in science fiction, such as 2001: A Space Odyssey and Star Trek, in which humans have conversations with computers and keyboards and mice are essentially obsolete.
The feature is available for version 27 of Chrome, which can be downloaded from Google.com or upgraded from an older version, provided your operating system is compatible.
Searches are activated by clicking a red microphone icon on the search page and speaking. At the I/O keynote, as seen on YouTube, Google's Johanna Wright demonstrated that Chrome not only understood her query and provided pictures of the Santa Cruz, Calif., boardwalk at her request, with a polite verbal response, but understood in a follow-up question "how far is it from here?" that she was still referring to Santa Cruz.
"This is a different way of parsing the question, and when you do it with voice it combines speech recognition with natural language search," said technology consultant Rob Enderle.
"It is still basically pattern matching, which simulates intelligence but it isn't intelligent. The system doesn't understand what you want, it guesses based on a number of factors and is most accurate when it knows where others reached successful results from saying something similar."
Enderle added that the system is intuitive to the user so it can make "assumptions" with increasing accuracy based on search history.
"For instance if you were to ask where the best place to eat is you might initially get a listing of Zagat-rated restaurants located worldwide ranked by rating," he told us. "But if it knows you are in San Jose, it might instead give you a local Yelp-ranked listing. But if it knows you like Mexican food, it might then adjust that listing prioritizing Mexican restaurants."
As yet, the system can't provide follow-up questions to refine the search, but eventually, Enderle predicted, the integration of social networking will allow for more personal recommendations based on a combination of more questions and established preferences.