Google is increasingly turning to artificial intelligence and machine learning to stay on top in web searches, the company has revealed.

Google scientist Greg Corrado told Bloomberg a large amount of the several millions of search queries entered by web browsers have been interpreted by the RankBrain AI system.
According to Corrado, RankBrain embeds large amounts of written language into mathematical entities interpretable by computers. This enables Google's search engine to offload unfamiliar search words and phrases to RankBrain, which can guess which other terms have similar meanings and present relevant results.
Google is investing heavily in artificial intelligence and machine learning, and RankBrain has become the third most important "signal" in the company's web search algorithm mix, Corrado said.
The difference between AI systems such as RankBrain and Google's existing "signals" that make up its algorithm for searches is that while the latter are based on discoveries and insights from developers, they do not learn.
RankBrain has been in operation for several months now, and Google believes the AI system is better than even experienced people at guessing which results should rank at the top for ambigious queries.
Search engineers at Google asked to predict the ranking of difficult queries got 70 percent of the answers right, whereas the AI system guessed correctly in 80 percent of cases.
Artificial intelligence and machine learning are rapidly developing, with giants such as IBM, Apple and Facebook all seeking to harness the technologies to gain a competitive edge.
Microsoft this year released its machine learning Project Oxford set of application programming interfaces and software development kits on open source code repository Github.
The company today announced the language understanding intelligent service (LUIS) public beta for Project Oxford. LUIS lets developers build models that understand natural language queries and commands for their applications.
ANZ Bank is currently working on an expert AI system, seeking to take advantage of the large amount of progress in computing technology, storage and networks over the past decades, as well as the big data volumes now available for analysis and interpretation.