Republic of Distrust

Big Tech Has Our Attention — Just Not Our Trust

For all of Silicon Valley’s huge innovations, two decades of broken promises and monopolistic misbehavior will require more than ChatGPT and faster chips to fix.

By Parmy Olson October 16, 2024

This column is a part of Republic of Distrust, a series about the loss of trust in American institutions and what can be done to restore it.

The tech industry likes to paint a fairy tale of living in the future, and true to form, its shimmering vision for artificial intelligence borders on the unbelievable. Sam Altman, the CEO of OpenAI, promises “shared prosperity to a degree that seems unimaginable” to humanity. Large language models that can draft emails and hold automated voice conversations will tutor our children, coordinate our medical care, shatter the frontiers of science and make human workers more productive. “Eventually,” Altman writes, “we can each have a personal AI team, full of virtual experts in different areas, working together to create anything we can imagine.”

A year and a half into the AI boom, the prosperity promised to humanity writ large remains mostly on the horizon. But in this modern-day gold rush, the people making the pickaxes are already benefiting handsomely. Since the launch of ChatGPT in November 2022, investors have added a staggering $8.2 trillion to the market valuations of tech’s Big Six firms: Alphabet Inc., Amazon.com Inc., Apple Inc., Meta Platforms Inc., Microsoft Corp. and Nvidia Corp. Buoyed by hype, Nvidia, a developer of AI chips, became the world’s biggest company by market capitalization in June. OpenAI has more than doubled its annualized revenue to $3.4 billion in the past six months. Microsoft’s revenue surged by 18% in the second quarter of 2024, driven by AI gains, to $62 billion.

There’s no denying the major advances that tech revolutions over the last few decades have brought to the world, from desktop computing to the internet to the rise of smartphones, often precipitated by visionaries like Steve Jobs, Bill Gates and the Google founders Sergey Brin and Larry Page. Their products have streamlined professional work and allowed businesses to operate on a global scale more efficiently, all while opening a continuously expanding universe of convenience, commerce and connection to billions of consumers. You don’t see many people rushing to give up their broadband connections and the powerful computers they have in their pockets.

Still, it’s not clear yet whether AI will benefit humanity as much as it will tech firms, and history offers ample reasons to be skeptical.

Facebook’s Mark Zuckerberg sold us on connecting the world but delivered algorithms rigged for endless scrolling, inadvertently straining mental health and political discourse. Google promised to organize information but allowed ads to dominate search results, while YouTube radicalized young men. Amazon’s Jeff Bezos created unparalleled convenience but forced countless smaller businesses into dependency or obsolescence. Harvesting the personal data of billions of people has fueled moneymaking across Google, Facebook and Amazon. Little wonder that Big Tech ranks below average in Gallup’s most recent polling on the level of confidence that the public assigns to US institutions.

Republic of Distrust

Read more from Republic of Distrust, a series about the loss of trust in American institutions and what can be done to restore it.

Yet despite consumers’ reservations, most are resigned to using tech platforms anyway. They’re our social infrastructure, mildly addictive and inescapably dominant in the marketplace. Leaders at the tech giants behind the revolutionary innovations of the Information Age have also come to see their ascendancy as deserved — and not subject to criticism. In September, when I pointed out to a senior scientist on Google’s search team that 90% of global searches were Google’s, he said the reason was not his employer’s anticompetitive behavior but smart choices by consumers. “That stat reflects that people trust Google,” he said. Well, sure. Google’s superior search algorithm got us on board, but so too did its acquisitions of potential competitors and its shrewd deals with device manufacturers: It paid $20 billion to Apple in 2022 alone to be the default search engine on iPhones. The story is similar at other Big Tech companies.

It’s telling that two visionaries who helped spark the recent AI boom, Altman and Google DeepMind co-founder Demis Hassabis, so distrusted the unbridled corporate control of AI that they tried and failed to structure their respective firms as nonprofits. In reality, the world cannot continue to trust Silicon Valley to self-regulate. Promising new efforts at policing tech are coming from the European Union, with the Artificial Intelligence Act and other laws tackling monopolistic behavior and social media harms. That is the current leading hope for balance.

From Smartphone Fairy Tale to AI Dystopia

The years after Steve Jobs’ launch of the iPhone in 2007 sparkled with optimism about how smartphones would enrich just about every aspect of our lives and drive progressive change, helping spread the word about everything from the Arab Spring to body positivity.

But for those watching closely, red flags started to flutter in 2014. That year, an online movement known as Gamergate orchestrated the harassment of women in the video-game industry, via large forums like Reddit. It turned out social media could be weaponized, too. Those concerns grew when Donald Trump was elected president in 2016, leading the technology press to question whether Facebook — with its new, algorithmic feed that rewarded outrage and engagement — had helped get him into office. As Twitter made similar tweaks to its own feed, the idea of filter bubbles started to take hold.

Two years later, Facebook admitted that it had let Cambridge Analytica, a consulting firm, collect the personal data of millions of its users for political advertising. So began Zuckerberg’s transformation to industry villain, a symbol of the unsavory business model that underpinned the internet giants: Advertisers were the customer, people and their data were the product. Website owners had been taught for years that they needed to make themselves sticky, but now consumers were realizing they didn’t always like getting “stuck” on their sites and apps, or having their brains conditioned to chase dopamine and clickbait. In 2021, a whistle-blower at Facebook sparked a seeming public reckoning for the firm, revealing that its own researchers believed Instagram, which Facebook owned, made teenagers feel worse about themselves.

This was heralded as tech’s “Big Tobacco moment.” But the swift reckoning that came for cigarette makers, which agreed to pay $200 billion in settlements just two years after major whistle-blower revelations in 1996 — hasn’t materialized for tech. Instead, the world’s six largest tech firms have surfed market excitement over their ability to capitalize on the pandemic and generative AI to reach unprecedented heights. Nvidia, Microsoft, Meta, Apple, Amazon and Alphabet dominate the S&P 500 more than any previous industry and have a combined market capitalization of $15.4 trillion.

The mistrust of tech’s oligopoly runs deep even in Silicon Valley, where the big companies’ not-so-giant peers and venture capitalists gripe behind closed doors about the industry’s kingmakers. Startup founders, for instance, must lock themselves into one of three leading cloud firms to host their computing and AI infrastructure: Amazon, Microsoft or Google, a trio that has also been hoarding access to the most powerful AI chips from Nvidia. All the while, founders and VCs must make nice with Big Tech executives since their likely endgame will be a sale to one of them.

Consumers are locked in a restrictive relationship, too. Microsoft software is used by 70% of humans with a computer, while Amazon accounts for 40% of e-commerce sales in the US. It’s almost impossible to go about daily life without the services of a tech giant, a phenomenon captured by writer Kashmir Hill, who spent six weeks in 2019 trying, and failing, to cut five of the top firms out of her life. Sending files for work, listening to music and getting information from the web all became “hell,” she wrote at the time. Despite the harmful side effects, tech giants are so deeply woven into our infrastructure and personal lives that consumers can’t avoid them. We cannot vote with our feet (or our clicks). The relationship has become as inescapable as the one with our parents: necessary, complex and largely nonnegotiable.

Perhaps that’s why public trust in Big Tech firms has been sliding. The number of people who have confidence in large technology companies stood at 27% in 2024, down from 32% in 2020, according to Gallup, despite the important roles that Zoom, Google Classroom and other products played during the pandemic.

The generative AI boom has done little to allay that skepticism. ChatGPT burst onto the scene, generating poetry and images and making Siri and Alexa look antique. OpenAI pitched it as a magical elixir to “empower humanity to maximally flourish.” As Google, Amazon and Meta scrambled to catch up, investors salivated over potential dot-com-era riches. But unlike the dot-com years’ focus on growth, generative AI’s humanitarian veneer masked a darker premise: This boom was about cost cutting, threatening to displace jobs rather than create them.

Silicon Valley’s approach to AI doesn’t inspire optimism that trust can be rebuilt. Sam Altman was fired by a board that believed he’d been lying about OpenAI’s philanthropic focus, before he fought his way back to power. Tech firms raced one another to scrape the work of artists and writers off the internet, rattling off half-baked software that told people to put glue on their pizza or that plagiarized copyrighted work. Google at one point was so desperate to look better than OpenAI that it released a video demo of its new chatbot that was egregiously edited to look more sophisticated. And as their corporate rivalry intensified, so did their secrecy. OpenAI refused to say what text and artwork it had taken to train its models, while Google’s DeepMind ordered its scientists to stop publishing their research without permission.

One result is that while consumers in the US and Europe believe AI will play an important role in their lives, they don’t trust it. Europeans especially worry that personal-assistant AIs will infringe on their privacy and manipulate their decisions.

Another result is that Americans don’t believe tech firms can regulate themselves. Around 65% of US consumers say they don’t trust the CEOs of companies that develop AI models to build their products responsibly and ethically — and they have even less faith in Congress to regulate those companies, according to a March 2023 survey by Morning Consult. Lawmakers had the survey’s worst result on the question of regulating AI.

In the last 20 years, tech companies have arguably proven themselves more influential than governments, creating multibillion-dollar industries from apps to the gig economy to the online advertising market, and inventing jaw-dropping products like ChatGPT.

In the US, regulators look hamstrung. The Supreme Court’s decision to kill the Chevron deference — a policymaking model that allowed Congress to set out broad rules for technically complex industries — has made the prospect of crafting federal laws to regulate AI even more daunting. Congress has yet to pass any legislation on social media or digital privacy, either.

Last year, the White House at least secured “voluntary commitments” from seven leading tech companies to better manage the risks posed by AI. It then followed up with an executive order applying to other tech firms and government departments. In laying out her policies should she win the upcoming presidential election, Vice President Kamala Harris has spoken more about managing the risks of AI than former President Donald Trump, who has pushed for lighter regulation on the technology and said he would rescind President Joe Biden’s executive order.

Tech firms in the meantime have done an excellent job of framing themselves as responsible stewards. AI companies say they probe their models for biases and flaws — a process known as red teaming. Facebook and Uber Technologies Inc. have publicly lauded their own “transparency reports” into harms that take place on their platforms, even though such surveys lack crucial data and aren’t audited by third parties. In reality, the continuing regulatory vacuum for software and the internet has allowed firms operating in those industries to cut corners in the pursuit of growth, especially now as they roll out new AI services.

Take, for example, the generative AI companies selling software for producing deepfakes. Few of them have paid for staff to monitor what their tools are churning out, according to my recent analysis of such companies. One result was that nefarious actors used HeyGen, an AI firm in Los Angeles that raised $65 million, to clone a woman’s likeness from YouTube and, to her horror, make her the face of hundreds of Russian- and Chinese-language video ads. HeyGen hasn’t responded to multiple requests for comment about that incident. With no regulatory oversight, responsible AI is voluntary. AI companies like San Francisco’s Anthropic, with $8.4 billion in funding from investors including Amazon and Google, can invest in red teams to test their systems for harmful side effects, but the hundreds of other startups that have jumped into this hot new market cannot afford such checks and aren’t obliged to.

Larger social media platforms, where such AI content is being shared, aren’t doing much better. Ahead of the French election in July, researchers found that the country’s political parties had made 81 posts across Facebook, Instagram and X featuring AI images, according to AI Forensics, a nonprofit group. Neither Meta nor X flagged any of that content, even though both had signed a voluntary pledge to tell voters when they were shown AI-generated political content. That should come as no surprise. Last year’s wave of tech layoffs hit Big Tech’s “trust and safety” teams hard. Google in early 2023, for instance, cut one-third of a division known as Jigsaw that worked on protecting users from misinformation and radicalization. (Google did not respond to multiple requests for comment on Jigsaw.)

Tracking how well tech firms are fighting hate speech or misinformation is a crapshoot. Research groups that do such work often operate on shoestring budgets. One of the biggest, the Stanford Internet Observatory, recently shut down. A saving grace for now is that there’s little evidence that the expansion of AI fakery is impacting elections this year: Mass persuasion campaigns are difficult to orchestrate, targeting particular audiences already drowning in information is likewise hard and voting behavior is determined by a range of complex factors.

Swipe Left on Self-Regulation

Still, history shows it’s a dangerous folly to rely on tech firms to self-regulate. Who can take up the slack? Not the US, where lawmakers have few incentives (read: campaign contributions) to do more than complain about Big Tech’s harms, which they will happily do, and who also face relentless lobbying by tech firms and super PACs advocating for the fields of crypto and AI. A few states have rolled out laws to protect local citizens from digital privacy infringements, but their officials are also juggling the prospect of new jobs for their voters as tech giants build out their data centers. Wall Street cannot be relied upon to push for change. One head of an independent equity research firm who despises Facebook for the toxicity it has bred in society tells me he still advises investors to plow money into the stock. It is printing money, after all.

Sans any European-based tech behemoths, the continent has naturally made the biggest strides in regulation, with two new laws that address both the tech giants’ market dominance and the harms to mental health posed by social media. Known as the Digital Markets Act and the Digital Services Act, both laws started applying to businesses in 2024; their full impact will unfold in the coming years as cases are brought forward. The rules in the EU’s AI Act, which focuses on potential harms, should start taking effect in mid-2025.

Until then, tech giants will benefit from our captured interest and the aura of respectability they cultivate with armies of public relations staff. Mark Zuckerberg says he has stopped apologizing for Facebook and has reinvented himself as a silver-chain-wearing maverick who can chug a beer, hoist the American flag and surf at the same time. Now Sam Altman is Big Tech’s villain du jour. While another scandal will no doubt engulf OpenAI, collective amnesia will set in among the public, just as it did with Facebook, and we’ll go back to scrolling our way around Silicon Valley’s delightful tools.

For years now, the public’s vacillating mistrust has not mattered to Big Tech, and for good reason: Its products are simply too necessary, too addictive, to boycott. That is an indictment of the systemic forces that allowed the industry’s dominance to grow unchecked. It also underscores the complexity of regulating companies that have become like utilities. Tobacco companies were obvious baddies, but we need Google and Amazon to run our businesses and send our emails. The result is quiet resignation and countless free passes, including over the AI-generated spam that’s overrun Facebook, the toxicity on Elon Musk’s X, the anticompetitive tactics from Amazon and Apple, the misleading results from Google, and more.

With little chance of a federal law to regulate Big Tech, and the EU’s own rules only just starting to take effect, those with political sway should speak up more about technology’s risks. In the UK, former Prime Minister Tony Blair has unhelpfully blown more hot air into the AI bubble. Former President Barack Obama, on the other hand, helped draft the White House’s AI policy and has talked up AI’s promise as well as the pitfalls that civil society campaigns have been warning about. Obama could put a bigger spotlight on those warnings. And Surgeon General Vivek Murthy could continue using his platform to warn Americans about the dangers of social media to mental health, as he has already done.

Another way to repair trust in tech’s incumbents may be to pay more attention to the industry’s more trustworthy actors. Anthropic, for instance, has given itself strict rules on privacy. While OpenAI collects its users’ prompts to ChatGPT — including personal ones — to train the chatbot further, Anthropic won’t do so without “explicit permission.” That puts Anthropic at a technical disadvantage to OpenAI, but the company still managed to launch a chatbot known as Claude 3.5 Sonnet earlier this year that bested ChatGPT in capabilities. Let’s hope Silicon Valley’s next generation of tech behemoths learn that unscrupulous tactics don’t always give them an edge, and that winning our trust will pay off in the long run.