It’s time for Sundar Pichai to step up and be more clear about Google’s A.I. search plans

Alphabet CEO Sundar Pichai
Alphabet CEO Sundar Pichai
Michael Short/Bloomberg via Getty Images

The generative A.I. revolution is taking place at such high speed that Google CEO Sundar Pichai’s statements on the implications can feel frustratingly incremental.

In a new interview with the Wall Street Journal, Pichai let it be known that Google Search will incorporate those newfangled capabilities. “Will people be able to ask questions to Google and engage with LLMs [large language models] in the context of search? Absolutely,” he said. 

Does that tell us almost nothing more than when, two months ago, Pichai told investors Google would at some point be integrating “more direct LLM-type experiences in Search”? Absolutely. 

Pichai is obviously aware of the pressure being placed on him and Google by Microsoft, which has already integrated OpenAI’s GPT technology into the front end of its Bing search engine. However, here’s what he told the New York Times a week ago: “You will see us be bold and ship things, but we are going to be very responsible in how we do it.”

That’s commendable in a way. If last week’s flap about the call for an A.I.-development moratorium has taught us anything, it’s that restraint is becoming a rare commodity in the tech industry. And Google has every reason to play it cautiously.

The main public demonstration of where the company is at with generative A.I. is its Bard chatbot, whose underlying LLM was recently upgraded from LaMDA to the larger PaLM model. Bard can easily be tricked into slipping out of its guardrails and spewing misinformation, the Center for Countering Digital Hate said in a report yesterday. Also, generative A.I. isn’t generally in the facts business, which is why Bard product lead Jack Krawczyk has been at pains to spell out that “Bard is not Search.”

“Bard and ChatGPT are large language models, not knowledge models. They are great at generating human-sounding text, but they are not good at ensuring their text is fact-based,” Krawczyk told employees a month back. “Why do we think the big first application should be Search, which at its heart is about finding true information?”

Indeed, Google Search has a 93% market share worldwide, and the company wouldn’t want to mess that up by making it less reliable. That kind of market share also puts Google in a pretty well-defended position—but here’s where I’m less sure about Pichai’s confidence. 

Pichai told the Journal that chatbots don’t threaten Google’s search business, as “the opportunity space, if anything, is bigger than before.” Maybe, but not for Google—when almost everyone is already using your search product, the only way to go is down. Bing, the market share of which is still under 3%, has nothing to lose and everything to gain by enthusiastically plowing into an A.I. future. I’m struggling to see how generative A.I. might grow the overall search pie, so Google really does need to have a strong (and yes, responsible) defensive game.

Here’s another Pichai quote (this time from last week’s NYT interview) that feels dangerously rose-tinted: 

“One of the things that gives me hope about A.I., like climate change, is it affects everyone. Over time, we live on one planet, and so these are both issues that have similar characteristics in the sense that you can’t unilaterally get safety in A.I. By definition, it affects everyone. So that tells me the collective will come over time to tackle all of this responsibly.”

From where I’m sitting, humanity is most definitely not solving the climate emergency. Yes, we’re making progress, but we are not making the urgent systemic changes we need to defuse what the U.N. calls a “climate time bomb.” In fact, the global incoherence in suitably addressing that crisis (and the COVID pandemic, for that matter) is what made me instantly dismiss that call for an A.I.-development moratorium as a well-meaning flight of fancy.

Crises don’t just sort themselves out. Google certainly shouldn’t rush into chatbot-ifying the Search interface for the sake of it, but if Pichai wants to provide reassurance that the company’s reluctance isn’t just down to cost implications or a lack of overarching strategy, at some point he is going to have to clearly spell out his plan.

Google’s annual developer conference takes place in May, and A.I. is expected to be front and center. Here’s hoping Pichai takes the opportunity to show leadership on an important issue that we all have a stake in.

Want to send thoughts or suggestions to Data Sheet? Drop a line here.

David Meyer

Data Sheet’s daily news section was written and curated by Andrea Guzman. 

NEWSWORTHY

VCs are spending big on generative A.I. The median pre-money valuation for generative A.I. startups hit $90 million this year, up from $42.5 million in 2022, PitchBook data provided to Fortune shows. The data is based on nine deals PitchBook tracked through March 29 and comes after big funding rounds in firms like Anthropic and Tome. Reflecting on the price tags, VCs say valuations are simultaneously justified (generative A.I. has the potential to disrupt entire business categories), yet also too high (only a select few of these A.I. startups will ultimately succeed).

Half the ads for Twitter Blue subscribers. Twitter users paying a monthly fee for the premium version will finally get a long-promised benefit: fewer sponsored tweets in their feeds. Twitter Blue subscribers will now see half as many ads as nonpaying users. The premium users will also get “prioritized rankings in conversations and search.” It remains to be seen whether these added benefits persuade more users to pay $8 per month for Twitter; data suggests that less than 1% of the platform’s monthly users are Twitter Blue subscribers, TechCrunch reports.

Samsung’s disappointing profits. South Korean chipmaker Samsung is expected to report that its quarterly operating profit dropped about 90% to $1.1 billion, its smallest profit since 2009. That’s according to analyst estimates compiled by Bloomberg, which reports that investors have called on Samsung to cut production of memory chips. Executive chairman Jay Y. Lee appears unlikely to follow that advice, telling the company’s executives earlier this year that they should not be fazed by the industry’s challenges. 

ON OUR FEED

“Last year, we likened it to ending the best party in town—but instead of simply turning on the lights, the past year has been more akin to getting cold water thrown in our faces.”

Billionaire venture capitalist Chamath Palihapitiya in his annual letter to Social Capital’s limited partners. Palihapitiya earned the nickname “SPAC king” for bringing a number of high-risk startups to market via his special purpose acquisition companies.

IN CASE YOU MISSED IT

Elon Musk’s brother Kimbal dumps $17 million worth of Tesla shares on the market, by Christiaan Hetzner

Three of Meta’s top execs—including Mark Zuckerberg—are now spending most of their time on A.I. in a bid to claw into market, by Eleanor Pringle

Twitter just designated NPR as ‘state-affiliated media’—and Elon Musk approved Twitter’s policy rewrite to intentionally change its status, by Tristan Bove

ChatGPT falsely accused a mayor of bribery when he was actually the whistleblower—now he wants to sue in what could be the first defamation case against a bot, by Prarthana Prakash

Stabbing death of tech executive unleashes torrent of anger from VCs and techies who ‘hate what San Francisco has become,’ by Ben Weiss

BEFORE YOU GO

Amazon’s bet on gems to boost computer networks. Artificial diamonds could help revolutionize computer networks, so a division of the mining company De Beers Group will be working with Amazon Web Services’ Center for Quantum Networking as it seeks to transmit data securely over longer distances.

The diamonds come into play since quantum networking uses subatomic matter to deliver data, and diamonds would help the data travel farther without degrading. Antia Lamas-Linares, who runs the Center for Quantum Networking, talked to Bloomberg about the project. “We want to make these networks for AWS,” she said, estimating it will be in use within “years rather than decades.”

This is the web version of Data Sheet, a daily newsletter on the business of tech. Sign up to get it delivered free to your inbox.