Bloomberg Businessweek (June 19, 2023)

Год выпуска: June 19, 2023

Автор: Bloomberg Businessweek

Жанр: Бизнес

Издательство: «Bloomberg Businessweek»

Формат: PDF (журнал на английском языке)

Качество: OCR

Количество страниц: 68

Can Al Save Silicon Valley?

The next big moment has arrived. We’re here to start sorting the early winners from the hype

In late May, 300 entrepreneurs, venture capitalists, journalists and assorted selfdescribed thought leaders crammed into Shack15, a stylish social club on the second floor of San Francisco’s Ferry Building, where most spoke in soaring terms about what they saw as the next gold rush. The gathering, dubbed a “Generative Al Meeting of the Minds,” would’ve been unthinkable during the pandemic and improbable earlier this year, when the city’s main obsessions often seemed to be car break-ins and retail store closures. The night had the feel of a religious revival. “Something is happening, something is cracking open,” said the evening’s host, futurist writer Peter Leyden, in the first of many upbeat speeches. Just as everyone “was talking about the demise of San Francisco, how everyone is leaving the Bay Area, how no one wants to live in California, how we are in doom loops—that’s exactly the time you know the place is right about to burst open in reinvention,” Leyden said to applause. The speech, the whole event, captured the feeling coursing through tech circles these days: Silicon Valley is back.

Only a few months ago, layoffs, cratering share prices and startup valuations, and the fallout from scandals at FTX and Silicon Valley Bank dominated headlines. It felt as if the entire industry was primed for a long retreat after years of parody and techlash. Now the talk of the town is artificial intelligence: chatbots such as ChatGPT and Google’s Bard, image creation tools Dall-E and Midjourney. There’s hope that tech might inflict yet another Jobsian dent in the universe.

“Nothing in the tech industry was really working—remote-work trends, people leaving, there was crypto, which got pushed out of the US. I think the mood was quite bad,” says Sam Altman, chief executive officer of startup OpenAI, which was knighted in January with a $10 billion investment from Microsoft Corp. “So the fact that Al happened here, and Al companies are in-person in a world of remote-first work, I think it totally pulled the center of gravity back to San Francisco.”

It was arguably eight researchers at Google who laid the groundwork for this reversal. In a seminal 2017 paper called “Attention Is All You Need,” they proposed a novel system for how machines might learn like humans, or at least seem to. The devices would consider long sequences of data, like chunks of text, and weigh each word in relation to what came before, while considering grammatical patterns. This idea wasn’t merely a fancy version of autocomplete; it represented a breakthrough in getting computers to better mimic both human reasoning and a facility with sequences such as language and computer code.

When the Google employees presented the paper at an Al conference in Southern California that December, researchers from OpenAI were there, taking notes. The first version of OpenAI’s GPT, which employed this architecture, was released the following summer. (GPT stands for “generative pretrained transformer.”) Subsequent versions were fed more information from the web, including Wikipedia entries, Reddit posts and newspaper articles. Then ChatGPT, based on a version of GPT-3 and released broadly last fall, dazzled with its conversational legerdemain, and the more advanced GPT-4 followed earlier this year. “It’s a game changer, a world changer,” says Oren Etzioni, a professor emeritus of computer science at the University of Washington, expressing an enthusiasm common now in Al academia. “We’re at the very beginning of this, and it is a very fast-moving phenomenon.”

The result is a frenzy not seen since the dot-com fever of the late 1990s. Tech CEOs are reorienting their companies toward Al and raising their stock prices simply by mentioning the subject on earnings calls. Venture capitalists are reevaluating their portfolios and piling into Al startups. JPMorgan Chase & Co. estimated that Al excitement drove 45% of this year’s gains in the S&P 500 through April. That excitement also led to a frank reassessment of previous trends, such as web3 and the metaverse, whose appeal to regular folks now seems thin or imagined next to the prospect of smarter machines answering bigger questions. Beneath the marketingspeak rests the potential for genuinely amazing advances, such as the recent Al-assisted discovery of a new type of antibiotic treatment for a drug-resistant superbug.

That so much of this energy remains concentrated in Silicon Valley, the constantly transforming region between San Francisco and San Jose, marks another twist. During the worst of Covid-19, evangelists for a tech diaspora tried mightily to brand cheaper, less pandemic-restricted cities. Larry Ellison moved Oracle Corp.’s headquarters to Austin in 2020, and Elon Musk followed a year later with Tesla Inc. Miami Mayor Francis Suarez earned a national profile almost entirely for promoting the city as a cryptocurrency hub. Valley expats tweeted incessantly and insufferably about the charms of Miami, or Los Angeles, or New York, or Puerto Rico.

The largest tech companies, however, remain in the Bay Area and its kin to the north, Seattle. Alphabet, Apple and Meta, Microsoft and Amazon.com have been hiring Al talent for years. As most of them cut payrolls and streamlined operations to account for the relative austerity that comes with higher interest rates, they’re indirectly creating new competitors, too. For example, Alphabet Inc.’s consolidation of its two Al divisions, Google Brain and DeepMind, “will cause people to leave, and the best ones start startups, because they’re much more mission-driven than paycheck-driven,” says Vinod Khosla, founder of Khosla Ventures and one of OpenAI’s earliest backers. “That’s the raw material we feed on and that, in five years, will look like OpenAI today.”

For San Francisco, the Al boom is a rare bright spot amid embarrassing national headlines about crime, homelessness and open-air drug markets that threatened to drain the city of its tech class. Ivan Porollo was one of those founders drawn to the Bay Area by the famed startup school Y Combinator. But he fled during the pandemic, first for New York, then LA, then Lisbon. In January he returned to start a business that promotes Al hackathons and conferences, taking a name from the informal designation for the city’s Hayes Valley district, where Al startups have clustered: Cerebral Valley. “Everyone went into the new year with this newfound energy around Al,” Porollo says.

Many of these evangelists seem to know one another and exult in their petty rivalries. In the cramped Palo Alto offices of Character.AI, a service that allows users to create chatbots with persistent simulated personality traits or to interact with representations of real and fictional people such as Musk and Harry Potter, a dozen employees sit shoulder to shoulder at enormous computer screens while a tiny Bichon Maltese wearing a blue checkered vest wanders the floor. “We invented a lot of this stuff. We taught OpenAI most of what they know,” scoffs co-founder Noam Shazeer, a co-author of Google’s seminal Al paper. Shazeer credits the Valley’s staying power to its willingness to quickly redirect energies and resources from old fads—say, crypto mining—to new ones, like running Al algorithms.

Previous Silicon Valley booms have ended with casualties: money lost, dreams dashed and masses of startups folding. Alexandr Wang, CEO of Scale Al in San Francisco, thinks there’s more profit potential for tools that make organizations smarter and more efficient. His seven-year-old company recently licensed Donovan, its “Al-powered decision-making platform,” to the US Army’s XVIII Airborne Corps, where it will digest thousands of situational updates and intelligence reports to help war planners make better decisions. But Wang acknowledges that the tech world is running the same old playbook, funding an excess of companies to find a few winners, which attracts plenty of pretenders and bad investments. “There are a lot of Al tourists pretending to be natives,” he says. “Ultimately they’re just selling vaporware.”

Al skeptics have good reasons to be skeptical. They seize on the questionable accuracy of the new tools; the risks for misinformation, deception and bias against women and people of color; and the fallout if the industry’s ambitious promises turn out to be empty. And they point to recent history. A decade ago the tech industry promised self-driving cars. Today there are a handful of driverless taxi pilot programs and improved cruise control systems in modern cars, but the sector has been marked by retrenchment and retreat. “If there’s a boom, there’s definitely going to be a bust,” says Mar Hicks, an associate professor of technology history at the Illinois Institute of Technology.

Avoiding an Al hangover might as well be the mission statement for this annual tech issue of Bloomberg Businessweek, which aims to serve as a guide to this new frontier and as a bit of a hazards map, too. We’ll explain how Microsoft’s blockbuster $10 billion investment in OpenAI positioned it to exploit the trend and get a leg up on rivals; how Al is reshaping Silicon Valley’s hiring practices; how Huawei and other wireless companies have tried to prepare for the next big fight; and how it pushed Nvidia Corp, to the brink of a $1 trillion market value. Alphabet CEO Sundar Pichai has the last word on helping his company meet the moment and guarding against an industrywide Al meltdown.

Back in San Francisco, many people aren’t ready to consider that the Al boom might amount to just another bubble. This, too, may explain the area’s remarkable endurance at the center of American tech. The young engineers and entrepreneurs flocking to Cerebral Valley and its environs weren’t necessarily around to learn the painful lessons from the last cycle about the fickleness of trends and the risks of competing against the resource-rich tech giants. They’re pleasantly naive, bursting with ideas, and ready to pick up and dust off a region and remake it entirely—at least until the next hype cycle hits.

Ok, Computer

• What the tech industry’s new obsession means for its fortunes—and yours

How Microsoft Took the Lead

• OpenAI’s largest shareholder is poised to reap billions

Nvincible

• Cutting-edge Al systems need Nvidia’s chips. There’s no way around that, yet

The Jobs Market: It’s Complicated

• Brutal layoffs may be over, but workers are left with tough questions

Huawei’s Danish Caper

• That time when the internet’s future seemed to come down to a mole hunt

IN BRIEF

• Inflation eases
• Ukraine strikes back
• Disney delays

OPINION

• The message of Miami: Even Trump isn’t above the law

AGENDA

• A tense Gay Pride
• US building permits
• Asteroid City

REMARKS

• Private credit is soaring, and regulators are taking note

BUSINESS

• Illumina faces new headaches as the CEO exits
• When AT&T says “RTO,” does it really mean “job cuts”?

TECHNOLOGY

• ByteDance’s bid to be all things to all people
• Spotify gives its podcasts a radical rethink

FINANCE

• The walls close in tighter on crypto
• What it all boils down to: Is a digital coin a security?

ECONOMICS

• Efforts to narrow the gender pay gap need teeth to work
• Xi’s popular tax-cutting policy may have to do a U-turn

SOLUTIONS / AVIATION

• For Boeing, the skies in China are clearing again
• A high-speed, low-emission, in-the-shop-a-lot engine
• Emirates and Qatar Airways have a cheeky new rival

LAST THING

• So what’s Google’s CEO going to do about all this Al?