A recently deployed artificial intelligence (AI) system is now helping to power a significant portion of Google searches. The AI system known as RankBrain is being used to generate relevant search results for unique queries, according to reports.
Such searches, which account for around 15 percent of the queries Google sees each day, can't be effectively handled by the previous results-based methods used to drive more familiar search queries, senior research scientist Greg Corrado told Bloomberg News.
Instead, RankBrain uses artificial intelligence to find the best possible results to unfamililiar search requests by embedding "vast amounts of written language into mathematical entities -- called vectors -- that the computer can understand."
Google uses a number of signals -- each weighted differently -- to determine the best results and ranking based on any particular search term. RankBrain currently ranks as the third-most important of those signals, Corrado said.
Growth in Mobile Search
Delivering relevant search results as quickly as possible has become increasingly important as more and more users conduct searches via mobile devices instead of desktops. In its most recent quarterly financial report, released last week, Google said it has seen "substantial growth" in mobile search revenues.
"With six products now having more than 1 billion users globally, we're excited about the opportunities ahead of Google, and across Alphabet," Ruth Porat, CFO of Alphabet, Google's new parent company, and Google, said upon the release of financial results for the third quarter of 2015.
Google researchers explore everything from machine learning and artificial intelligence to human-computer interaction as well as speech and natural language processing while working to develop and improve the company's products.
"Research at Google is unique," according to the company's research site. "Because so much of what we do hasn't been done before, the lines between research and development are often very blurred."
'Rethinking Everything We Are Doing'
Google last month revealed that it had started using "deep neural networks" to improve the quality of its voice-based search results. The new approach enables computers to more accurately "hear" spoken words on the fly in real time, the Google Speech team noted in a blog post.
Ongoing advances in machine learning and AI are being used by Google and many other tech companies to make their systems faster, smarter and more responsive, even as the volumes of data they must handle expand dramatically.
"Contrary to much of current theory and practice, the statistics of the data we observe shifts very rapidly, the features of interest change as well, and the volume of data often precludes the use of standard single-machine training algorithms," Google said on its research page for artificial intelligence and machine learning. "When learning systems are placed at the core of interactive services in a rapidly changing and sometimes adversarial environment, statistical models need to be combined with ideas from control and game theory, for example when using learning in auction algorithms."
During last week's financial earnings call, Google CEO Sundar Pichai said, "Machine learning is a core transformative way by which we are rethinking everything we are doing."