Why Google's Artificial Intelligence Boss Is Taking Over the Search Empire

As it ballooned, Google’s research group has nabbed a shocking number of computing’s biggest brains. Geoffrey Hinton. Peter Norvig. Ray Kurzweil. Titans of the field. And it held onto its homegrown talent younger minds, like Jeff Dean, a fabled technician.

All of them worked the tech and research circuits, got their faces out there. But all of them work for a much lower-key engineer: John Giannandrea.

Starting next month, the entire search organization — the beating heart of Google’s $75 billion colossus — will work for him, too.

On Wednesday, Google’s veteran search chief, Amit Singhal, announced his retirement from the company. Rather than tapping one of his deputies, Google put Giannandrea, who runs its sprawling research division, in his place and merged the two divisions.

The shift caps a broader trend, as machine intelligence advances have crept into Google’s core product. (Google’s founders are, after all, artificial intelligence nerds.) And it reveals the company’s thoughts on the future of search, which is moving to places that need smart AI, like voice, and is something Google must innovate upon and control. It is especially critical on smartphones, where users prefer apps over the Web and which Google doesn’t dominate the way it does desktops.

The retiring Singhal, who came to Google from AT&T’s Bell Labs in 2000, is part of a small circle of longtime execs with influence over core products at Google. He is respected inside Google and in the world of Web search and is known as a fierce defender of the sanctity of results (and his search domain). Yet Singhal is not known well outside the company.

Adam Berry / Getty Images

Even fewer know Giannandrea. But the mild-mannered engineer — described, like many of CEO Sundar Pichai’s deputies, as lacking a big ego — is no novice. He cut his teeth at pioneering Internet companies, and since joining Google in 2010 has overseen the herculean task of moving some of its more far-fetched research from the lab into actual consumer products.

“He’s an OG Valley technology visionary,” said Marc Andreessen, co-founder of Netscape, where Giannandrea worked as chief technologist during its peak growth. Before that, Giannandrea had stints at early tech firms Silicon Graphics and General Magic.

“[He’s] always on the leading edge; [I] always learn new things when I talk to him,” Andreessen continued in an email. “Great news for Google.”

Google declined to make Giannandrea available for an interview. But back in the fall, he opened a small information session on machine learning for reporters at Google’s campus. He spoke softly but with a clear command of the material, walking through the significance of the field for Google, very briefly, before handing the stage to his lieutenants.

“Recently, it’s been taking an increasingly important role,” Giannandrea said then about the AI approach. “We think that something really big is happening — something new and significant.”

The Machine in the Results

That something new is a renaissance in the subfield of artificial intelligence — one where Google is ahead, albeit with ambitious rivals behind it.

Machine learning is, in brief, a method for computers to process data on their own, without being explicitly programmed. Thanks to hardware advances and the now gigantic corpuses of data tech companies have, the decades-old science is rapidly entering the real world.

For instance, it’s coming to search.

From its birth, Google’s search algorithm has been advanced — a secret sauce of weights to surface results. It was also coded by human hands. For some years, Singhal was reluctant to bake machine learning into results, according to former Googlers.

But it has slowly crept in, beginning with instant search results and the “knowledge graph” — a connective web of related things that Giannandrea helped bring to Google. Then last year, the company rolled out an AI, dubbed RankBrain, into the search sauce, as Bloomberg first reported. It can, among other things, process never before seen searches with more efficiency.

This is key in keeping Google’s engine from growing stale. It also enables more predictive search, delivering results with alarming accuracy. Google cracked intent, overhauling a decade-plus of search engine optimization practices in a short time, said Rand Fishkin of Moz, a marketing software firm.

Deciphering first-time searches is critical too because more future queries will come on mobile devices in languages other than English, a point Giannandrea made at the session in the fall.

He came to Google in 2010 with the acquisition of Metaweb Technologies, where he was CTO, working on a different variant of AI — one based on connecting related things automatically. Google does this, along with other AI approaches, but it has embraced machine learning systems, like the one behind RankBrain, more warmly.

“They allow for much more sophisticated and accurate classifications and predictions about data,” said Chris Nicholson, co-founder of deep learning startup Skymind. “That’s important for Google, because they want to make more accurate predictions about what people are searching for, and what they’re going to buy.”

The Search Fortress

To do that, Google has to be a part of the search. Getting cut out of that process remains one of the gravest concerns at the company. Despite still raking in search ad money, Google faces the threat of losing mobile searches to apps and rivals Amazon and Apple, which have their own machine learning efforts.

Hence the rapid evolution away from the old 10 blue links toward Google’s own content, particularly on mobile. Hence the aggressive push to index mobile app content for search, which Singhal led. And hence the absorption of Google Now, the company’s AI-enabled personal assistant, into the search group.

While key to Google’s search strategy, that shift created some ripples inside the company. It’s now the type of politics Giannandrea has to handle. A former Googler sang his praise, suggesting that he may be able to change entrenched, “religious” practices in the search org.

If he does, he’ll have to do so while also managing the growing team of very smart researchers, ensuring that their work is relevant for Google. And he’ll have to keep the magic alive in that unit, which is competing for talent with several tech giants.

“We actually don’t know what the future of our science is going to be,” Giannandrea said at the fall event. “Once it starts working, people don’t think it’s impressive anymore.”

This article originally appeared on Recode.net.

Will you support Vox’s explanatory journalism?

Most news outlets make their money through advertising or subscriptions. But when it comes to what we’re trying to do at Vox, there are a couple of big issues with relying on ads and subscriptions to keep the lights on:

First, advertising dollars go up and down with the economy. We often only know a few months out what our advertising revenue will be, which makes it hard to plan ahead.

Second, we’re not in the subscriptions business. Vox is here to help everyone understand the complex issues shaping the world — not just the people who can afford to pay for a subscription. We believe that’s an important part of building a more equal society. And we can’t do that if we have a paywall.

So even though advertising is still our biggest source of revenue, we also seek grants and reader support. (And no matter how our work is funded, we have strict guidelines on editorial independence.)

If you also believe that everyone deserves access to trusted high-quality information, will you make a gift to Vox today? Any amount helps.

One-Time Monthly Annual

$5/month

$10/month

$25/month

$50/month

Other

$

Yes, I'll give $5/month

Yes, I'll give $5/month

We accept credit card, Apple Pay, and Google Pay. You can also contribute via