The search engine is the heart of the web experience and drives many of its revenue models. Hence the power of Google, and the huge efforts that rivals are making to try to wrest control of the next generation of search away from the leader. Apple made important baby steps with Siri, which brought voice activation into the mainstream and was a step towards the first important shift ‘ to provided a complex but direct answer to a question, not a list of results. Google hit back at Siri with Google Now, but CEO Larry Page acknowledged in his annual ‘Founder’s Letter’ that ‘in many ways, we’re a million miles away from creating the search engine of my dreams ‘ the one that gets you just the right information at the exact moment you need it with almost no effort’.
The quest for that engine is involving a broadening range of emerging technologies, from new inputs ‘ voice, gesture and facial ‘ to new ways to present the answers, including virtual reality. Most importantly, it is harnessing the kind of deep machine learning and artificial intelligence that was once confined to sci-fi movies and defense laboratories. ‘Information is Google’s core,’ as Page noted in his letter ‘ gathering and analyzing vast quantities in all kinds of formats, in order to deliver the new breed of search results, but also to track people and machines more efficiently than ever before, to power the core revenue generators, advertising and big data.
In this area, Google has some powerful challengers, most of them its usual web rivals ‘ Facebook, Microsoft, Amazon, Apple ‘ but also some with less obvious business models for their advanced AI platforms, none more fascinating than IBM with its Watson technology. The pace of change in big data and personalized, context-driven consumer services means that some of the changes which Google has invested mightily to achieve already look pedestrian ‘ like voice search, which now works in 38 languages according to Page.
Yet this is an important step to making search ‘conversational’. Page wrote: “Improved context will also help make search more natural, and not a series of keywords you artificially type into a computer. We’re getting closer: ask how tall the Eiffel Tower is, and then when ‘it’ was built. By understanding what ‘it’ means in different contexts, we can make search conversational.”
The vastly complex learning engines which the cloud giants are building will initially be used mainly to improve the search experience and therefore drive new uses of the web, increasingly on mobile devices from phones to cars and wearables. Here Google has the significant advantage, over Facebook in particular, of Android, both as a channel for its applications to a vast base of gadgets, and as a very simple way to access those services.
But the back end engines will only earn their living if they drive entirely new revenue streams too. For Google, Microsoft and IBM, probably the biggest investors in this kind of platform, they can drive new efficiency and new added value into basic cloud services such as data analytics or storage.
IBM, for instance, has just announced Elastic Storage, which is driven by Watson. As with virtualized servers, it allows enterprises to access any data from many storage locations, including in the cloud,.in a unified way. The AI capability comes in for intelligently moving the most-used data to the fastest, but most expensive storage medium, such as flash, a process which can reduce storage costs by 90% for the enterprise or cloud provider, says IBM. When Watson famously appeared on the game show Jeopardy in 2011, defeating two human contestants, 200m pages or 5Tbytes of data were loaded into its memory in the space of minutes.
This is just the latest example of IBM harnessing one of its most famous R&D projects, Watson, as a weapon in its battle to offset declining hardware revenues with a new roadmap, laid down by CEO Ginni Rometty, centered on data analytics, cloud computing and mobile services. ‘We’re transitioning to key growth areas and transforming parts of the business,’ CFO Martin Schroeter told analysts. ‘We’re continuing to make investments in key growth areas such as mobility, security and cloud, and these initiatives are gaining traction but are not yet at scale.’
At this year’s Mobile World Congress, Rometty announced a global mobile developer challenge for Watson, encouraging business and consumer applications which require the ability to answer complex questions asked in natural language. She highlighted how Watson, originally created in IBM R&D labs to show what was possible when cognitive computing and natural language processing were combined, has now become commercially necessary technology. A whole division has now been created to bring Watson services to market, and build an important place for IBM behind the new mobile behaviors. The technology is already in real world use in a few applications such as complex healthcare diagnostics and despite the AI moves of Google and others, Rometty remains credibly confident about her firm’s headstart. “Every major invention in data and analytics has come from IBM,” she said.
Some recent moves include an investment in Fluid, a digital commerce firm, to develop personalized shopping applications using Watson. These would seek to provide the advice usually given by a store salesperson, to a mobile or online buyer. Watson is a significant element of a new start-up accelerator recently established by IBM Israel. And the supercomputer on which it runs has been made available to mobile developers, in order to expand the ecosystem and make IBM’s AI engine part of the mainstream, as well as feeding innovations back into the firm to help with its urgent bid to become a mobile software powerhouse.
The next stage, according to Moneshia zu Eltz, managing director of M&A and strategic investments at IBM’s Watson Group, is to develop algorithms that will enable the engine to learn on its own, a deep shift away from the traditional programming-based approach. ‘We are only beginning to explore how algorithms will be able to take big data, analyze it, and come up with something new,” zu Eltz told an event in Israel. “Deep natural language processing of the type Watson does is not keyword search.”
Of course, few companies have the deep R&D pockets of IBM or Google. Others will turn to the emerging providers of deep learning and AI ‘as a service’, and indeed, IBM and Google could themselves enter that field. An example of this new addition to the ‘aaS’ picture is natural language start-up AlchemyAPI, whose latest neural network system, AlchemyVision, is focused on image recognition.
Operators, too, are increasingly aware that their traditional revenues are drying up, but they still have considerable assets to leverage, not least their wealth of data about their customers’ behavior, from movements round the network to mobile purchases. Some are developing complex engines to analyze all this and could make these the basis of services to enterprises. AT&T Labs recently won the top prize at the 2014 Edison Awards, in the research and business optimization category, for its nanocube, which uses visualization technology to help users interpret massive sets of data in real time. The system was devised to help the carrier understand its own data, but could be applied to any large dataset, according to John Donovan, the firm’s senior EVP in charge of technology. “Users can create models connecting billions of data points, or quickly zoom into smaller sets. And unlike other visualization tools, nanocubes don’t require a supercomputer. Visualizations can be done within a web browser,” he wrote in a blog post.
Like IBM, AT&T will release some of its technology to open source in order to encourage others to build on it, speeding innovation and creating an ecosystem.
Microsoft will not be far behind, in an area where it has always been a heavy R&D investor. It recently set up a new special projects group within its R&D organization, according to insiders, which aims to take on the similar Google X unit, and is headed by Norman Whitaker, formerly deputy director of innovation at the US Defense Department’s research arm DARPA. The new Microsoft initiative is working on “disruptive technologies that could benefit the company and society’ and has a long list of targets, including machine learning/AI, mobility, big data, distributed computing and user experience design, according to a recent job advert.
While Whitaker (an expert in automated vehicles) looks over the horizon, Microsoft is building more immediate advances into its Bing search engine and Cortana personal assistant, which like Google Now are becoming increasingly AI-driven. Recently, it added a predictions feature to Bing, which analyzes inputs such as Twitter feeds and searches to forecast the results of voting shows like American Idol. It’s a trivial application, but one powered by a sophisticated engine, and Microsoft is vying with Google to lead in object and image recognition and analysis too.
So what of Apple? Its Safari search is still powered by Google but it is developing AI capabilities to support Siri and other web services such as the gradually improving Maps. Siri search relies on Bing, ironically enough, but it is highly unlikely Apple would promote Bing to the primary search box on the browser, even to reduce dependence on Google. There have been repeated rumors that it will turn to Yahoo, which at least does not have competitive hardware to the iPhone. But with search, in its various guises, still at the heart of the mobile and web experience, surely it would make sense for Apple to develop its own? It has the infrastructure to power a search engine and could hire the talent to create something cutting edge, and then unify it with Siri. Of course, given the disaster of the iOS Maps launch, Apple may not want to take the risk, but without its own platform, it risks remaining dependent on its fiercest rivals, as it has been in key iOS services.