Creepy or Helpful? Google Now and the Future of the Predictive Search

As if Google doesn’t already know enough about us.

Google is currently unveiling a new technology that promises to spoon-feed us answers before we even ask the questions. It will wake us up early for that morning meeting because it knows about delays on the train. It will tell us to bring an umbrella for afternoon rains. When we run by the supermarket after work, it will pull up the shopping list. Before I forget, it will remind me to call my mother on her birthday.

Part personal assistant, part psychic, Google “Now” silently mines our emails, calendar appointments, and phone calls. It looks at our location, search queries, and past behavior. Without us having to make a move, Now keeps us one step ahead. According to Google, Now gives us “Just enough information, at just the right time.”

At Google’s showcase of technology in San Francisco this May, Larry Page told a crowd of 6,000 developers, “Technology should do the hard work so people can get on doing the things that make them happiest in life.” Google Now simplifies many tasks by retrieving data the moment it is needed, preventing us from hitting heavy traffic or a downpour. Cool? Absolutely. Creepy? Almost certainly.

Technology that does more than crunch numbers, but actually monitors our habits can easily slide over from useful to unsettling. A certain dosage of creepiness—a combination of intrusiveness and, in the case of technology, personal forecasting— risks alienating users. This creepiness is inevitable because its actually how Google Now works.

Predictive searches use machine learning to forecast future behaviors by watching current behaviors. After monitoring habits, the program begins to spoon-feed users “cards.” These cards contain snippets of information in a certain category—stocks or sports, for instance—tailored to a user’s portfolio or home team. Users can shape the program by swiping away unwanted cards.  The more users tinker with Now, the more personalized the service becomes, and the more it unveils how much it actually knows.

The newest Android “Jelly Bean” phones already have Google Now preinstalled. For now, its travel functions are most compelling, updating users on flight delays, hotel reservations, local attractions, and currency exchange rates.

The anticipated release of Android 4.4, “Kit Kat,” will invariably demonstrate ever more unconventional uses for predictive searches. For instance, Google recently announced a new program that can comb through photo libraries, sift out blurry shots, and identify close friends and family. In other words, in a few months, Google will be able to spot your mother in a crowd. Less alarmingly, it may one day speak to other technology systems, automatically arming our home security system as we leave the house and reminding us that the milk in the fridge has soured. The serendipitous law of unintended consequences means we have yet to know what its most novel uses.

Rudimentary forms of predicative technology have been simplifying searches for the last decade. In 2004, Google launched Suggest—renamed AutoComplete in 2010—which generated instant search suggestions based on input, with both helpful and comic results. Typing in “why are…” for instance, immediately prompted two suggestions: “why is the sky blue?” and “why do men cheat?” Google launched Instant Search in 2010, a program that rapidly generated search results under the text during the typing process.

Google’s latest projects, including Glass and ChromeBook, have occupied the limelight. However, Now represents not simply a new technology, but a paradigm shift in our relationship to technology. Computers primarily digest inputted information. Using big data and smart algorithms, they can now speak back.

Moving from conventional user-generated searches to dynamic programs that anticipate future behaviors requires massive amounts of instantly intelligible data—big data as only big-brother Google can provide. Google Now is one of the first programs requiring cross-company collaboration. Indeed, various Google offices gather and integrate information: A traffic card enlists the Maps, Traffic, Calendar, and Text divisions; A “Photo Spots Nearby” card requires GPS, Images, and social networking departments to work in tandem. However, as personalized big-data collates, compiles, and consumes our information, privacy becomes a far more pressing issue.

The time is ripe for robotic assistants authorized to sift through our mounting piles of data and responsibilities and organize our lives. Juggling tens of appointments and hundreds of emails and text messages every day has proven difficult, if not impossible; we will forget to schedule a meeting mentioned in the fourth round of a group email; we will lose track of time. Everyone could use a social secretary. The only requirement is that, like a secretary, Google know just about everything happening in your life.

Google stores all our search information when we log into its email programs or use its Chrome internet browser. Google already has every Gmail we’ve ever sent or received. It has stored all our Gchats, Google alerts, calendar events since day one, and quite a lot of our browsing history too. It has extracted our data using powerful algorithms and feeds us customized data, news, and especially advertisements for years. It forever knows when we search for an individual, a medical symptom, or a political idea. It even stores our gossip.

Using Google Now, like much of Google, is a tradeoff. Surrender more privacy, and we are offered a litany of goodies in return. Refuse, and we are back to the age of Microsoft’s antiquated Internet Explorer. Google has tactically set up its services such that, while opting out is an option, few other options exist for privacy-inclined users. We want to the services, so we forgo the privacy.

While Google famously and unofficially assures us that it will “do no evil,” there is room to be skeptical, if not outright alarmed. Now that it has compiled data on millions of individuals, data can—and regularly does—slip into the wrong hands. A breach in Google Now’s security could exacerbate leaks from simple but recoverable social security numbers and credit cards, as have leaked in the past, to the entire identity of an individual: locations, conversations, even daily schedules.

Even if Google delivers unprecedented data security, government agencies, say, the NSA, could surreptitiously force its hand into delivering extraordinary detailed records of users.

Google Now requires users to perform a leap of faith, to believe that Google will act out of the best intentions. While the company has kept the public’s trust, offering services in return for targeted advertising; Google may one day turn more sinister. Conspiracy theories aside, Google is a profit-driven company, and if the end goal is capital, it may compromise our privacy or trust by selling our prized data that we trust are protected by passwords. For Google Now to succeed, it will have to ensure their our that data is not only secure, but does not compromise our confidence.

With privacy a side concern, Google Now can already look through our lives and do the tedious work of recalling locations and setting reminders. As a basic form of Artificial Intelligence, it can begin automating our lives in radical and unpredictable new ways. But as this technology rapidly improves, as it no doubt will in just a few months, when, if ever, will this mobile search tool stop acting like a tour guide and start acting like a stalker? When we will feel unease with Google recognizing our family members, favorite restaurants, and daily commute? When will Google Now plunge into the Uncanny Valley, the point at which our technologies start acting human and we start feeling uncomfortable?

I would like to suggest that the uncanny moment will never actually occur; in another creepy way, Google is one step ahead. The relatively limited number of cards announced in the first round of Google Now demonstrates a controlled release of this potentially revolutionary technology. We will not become psychologically accustomed to powerful predictive technology with the next software update. We need to become conditioned, or maybe worse, desensitized to invasive technology. We need to acclimate slowly.

The limits and most powerful uses for Now are still unknown, and our relationship to these technologies is yet to be determined. Google’s data empire has allowed it to explore an entirely new way to access data. Whether it will continue to do so depends on the care it shows in delivering data securely while balancing utility and privacy. One outcome is already certain: Now is only the beginning of the future.