Among all the AI helpers from major tech companies, Google Assistant is the most mysterious.
Google has billed Assistant as one component in "the next evolution of Google." Others, such as Danny Sullivan at Search Engine Land, have gone a step further, imploring us to think of Assistant as "Google 2.0." Backchannel’s Steven Levy has documented how Google is rebuilding its entire company around the type of machine learning that Assistant will showcase.
Yet despite all of Assistant’s potential, we know very little about how it’ll operate. Beyond the fact that Assistant will first live inside two products—a messaging app called Allo (due this summer), and a smart speaker called Google Home (release date TBA)—Google has left much to the imagination.
Two months after Google announced Assistant, here’s what I’m still scratching my head over:
Beyond just serving up information from the open web, Google has casted Assistant as a way to get things done. But so far, Google Assistant hasn’t demonstrated much more than the ability to look up restaurants and book them on OpenTable. Everything else appears to be aspirational.
At Google I/O, for instance, CEO Sundar Pichai walked through a scenario where he’d ask what movies are playing, and narrow down the genre by saying, "What if I bring the kids?" Unprompted, the Assistant would then offer to buy the exact number of tickets for Pichai’s family.
This magical interaction is the stuff of sci-fi, but it’s also "entirely hypothetical," according to a Google representative, who clarified to BuzzFeed that the movie ticket scenario was meant to demonstrate "how we think an assistant should work over time."
To make Assistant more useful, Google wants to hook into third-party apps and services. But Scott Huffman, Google’s vice president of engineering for search, told The Verge that the company is "trying to take our time and be pretty thoughtful" about the execution.
As we’ve seen with other virtual assistants, third-party integrations can play out in a lot of ways. Siri will only work within a narrow set of app categories, with Apple tightly controlling the conversation, while Amazon’s Alexa employs a more open system that lets developers create their own dialog trees. On the text messaging side, Facebook views third-party chatbots as distinct personalities that users can talk to, while Apple will let users summon apps through existing conversations with other people.
It's hard to argue that one of these approaches is clearly better than the others. Google's statements about moving with caution suggest the company hasn't yet figured out the ideal approach for itself, either.
Here’s an interesting question brought up by Joe Youngblood at Search Engine Land: Today, when you search for a specific fact in Google, like the number of stars in the galaxy or the release date of an upcoming video game, Google may give you a direct answer instead of just links. The source of that information typically comes from a web page, such as Space.com or Wikipedia.
As Youngblood notes, publishers may be okay with the practice today because of the click-throughs that this top search placement promotes. But what happens when the interaction moves farther away from a web browser, and onto hands-free devices like smart speakers and smartwatches? Does this cross a line where Google is taking too much, and not giving enough back?
Keep in mind, Google does allow websites to opt out of data scraping without affecting traditional search rankings. But as those blue links become less important, what choice will other web services have?
Google hasn’t gone into detail on how it’ll make money from Assistant, but Pichai has hinted at some kind of paid placement scheme for third-party services. "Inherently, when information needs are commercial, you are connecting users with people who provide services, I think there are natural opportunities," Pichai told Forbes in May.
This raises a new dilemma for Google: With web search, you might see sponsored listings at the top of search results, but they’re visually distinct from the organic list of blue links. In a system where Google isn’t just giving you information, but helping you quickly take action, what sort of advantages will paid services have over non-paying ones? Put another way, will more of our decision-making through Google be influenced by the highest bidder?
Although Google Assistant is supposed to be "conversational," it’s unclear how often the AI will be initiating the dialogue.
Proactivity, after all, has been a cornerstone of Google Now, which can alert you to sports scores, incoming packages, travel delays, and upcoming appointments through the Google app on smartphones. Still, these alerts aren’t always helpful, and are sometimes just annoying (like when you’re routinely shown directions home from familiar places, or offered more information about something you searched for once).
Google isn’t really talking about those scenarios with Assistant, but does note that Assistant will chime in with relevant information during conversations with other people. Does that mean Assistant is a clean slate for proactivity, or will it keep building off Google Now?
At first, Google Assistant will appear in two products. Allo is a messaging app, in which you can talk to your friends and to Google, either together or separately. Google Home, meanwhile, is a voice-activated speaker, similar to Amazon’s Echo.
Where will Assistant show up next? There are already rumors of new Android Wear smartwatches with Assistant as the centerpiece, and it’ll likely appear in more apps and hardware devices over time. But the bigger question is whether Assistant will ultimately replace the traditional search box on phones, tablets, and desktop computers. If Assistant is really the start of a new Google, we can only assume that’s the goal.
Perhaps the most important thing to know about Assistant it’s not exactly a new product. Many of its capabilities, like its understanding of natural language and follow-up questions, and its ability to dig through the web’s contents to answer those questions, have been kicking around in Google products for years. Not everyone seems to have realized this: When Google showed a video highlighting the capabilities of the current Google app, several tech sites claimed that it was a demonstration of Assistant in action.
The point of Google Assistant, then, is to insert Google’s existing skills into places where they previously weren’t available, like a messaging window or a connected speaker. In these new contexts, we’ll expect Google to do more than it does now, which may explain why the company has indulged in some fantasies about what Assistant might do in the future.
Assistant isn’t so much a product as it is a mission statement. As for whether it’ll succeed, we’ll just have to wait and see.