Tuesday, August 26, 2008

Google working on search engine with built-in artificial intelligence


Hyderabad June 22 If innovation and research work in Google's integrated globally dispersed R&D centres are anything to go by, your search experience would only get a lot better, more precise and possibly more intuitive with artificial intelligence built into it.

This means, when you search the Web, the engine would be able to identify a user's preference and throw up information specific to him.

May be some day, in the near future, you would be able to ask your search engine "What I can do for the day?" And based on the location and your choices, chances are that it will guide you. A perfect search engine will tell you exactly what you want and what you need.

"An ideal scenario is one that extracts information from multiple languages and gives the right information to a Googler. However, this scenario is still some time away as the search engine capability is barely scratched," says Mr Kannan Pashupathy, Director, International Operations, Google, who operates out of Mountain View office of Google.

During his visit to the India centre, Mr Kannan said that a few years ago, most of the revenues for Google were generated out of the US. As it expands and penetrates new geographies, Google has set up centres in countries such as India, China, Norway, Japan, Russia, Korea and other markets. Soon India and China would mirror this.

This called for lot of innovation from a local stand point. Even though the research work is virtually handled across Google's development centres, localisation would be the differentiator.

Google's research teams typically comprise five or six designers who work with a corporate mandate. However, they have what is referred to as "20 per cent innovation time," where they get to work on projects that they would like to. Such projects have thrown up Gmail and Google News, Mr Kannan said.

HARDWARE

Google is also an innovator in hardware.

Mr Kannan said that all of Google hardware is designed and custom developed in-house, including for its large data-intensive centres. The learning is such that Google could even develop high performance computers.

"All this was initiated as we felt that the hardware offered by others was costly. Our infrastructure is such that it is one of a kind giant supercomputer," he explained.

"The headcount of Indian designers based at Bangalore and Hyderabad has quadrupled in last year and we are poised to double this number. We continue to expand and recruit as and when we find talent. It is just not a numbers game," he explained.

Improving user interface, parallel computing, machine learning and distributed services are areas of focus and Google has emerged as a giant research engine. This is because the innovation spans machine learning, computer science, algorithms and statistical methods. Collaboration is key to our success, he explained.

courtest :Hindu