Itemoids

CEO

Was Sam Altman Right About the Job Market?

The Atlantic

www.theatlantic.com › technology › archive › 2025 › 03 › generative-ai-agents › 682050

The automated future just lurched a few steps closer. Over the past few weeks, nearly all of the major AI firms—OpenAI, Anthropic, Google, xAI, Amazon, Microsoft, and Perplexity, among others—have announced new products that are focused not on answering questions or making their human users somewhat more efficient, but on completing tasks themselves. They are being pitched for their ability to “reason” as people do and serve as “agents” that will eventually carry out complex work from start to finish.

Humans will still nudge these models along, of course, but they are engineered to help fewer people do the work of many. Last month, Anthropic launched Claude Code, a coding program that can do much of a human software developer’s job but far faster, “reducing development time and overhead.” The program actively participates in the way that a colleague would, writing and deploying code, among other things. Google now has a widely available “workhorse model,” and three separate AI companies have products named Deep Research, all of which quickly gather and synthesize huge amounts of information on a user’s behalf. OpenAI touts its version’s ability to “complete multi-step research tasks for you” and accomplish “in tens of minutes what would take a human many hours.”

AI companies have long been building and benefiting from the narrative that their products will eventually be able to automate major projects for their users, displacing jobs and perhaps even entire professions or sectors of society. As early as 2016, Sam Altman, who had recently co-founded OpenAI, wrote in a blog post that “as technology continues to eliminate traditional jobs,” new economic models might be necessary, such as a universal basic income; he has warned repeatedly since then that AI will disrupt the labor market, telling my colleague Ross Andersen in 2023 that “jobs are definitely going to go away, full stop.”

Despite the foreboding nature of these comments, they have remained firmly in the realm of speculation. Two years ago, ChatGPT couldn’t perform basic arithmetic, and critics have long harped on the technology’s biases and mythomania. Chatbots and AI-powered image generators became known for helping kids cheat on homework and flooding the web with low-grade content. Meaningful applications quickly emerged in some professions—coding, fielding customer-service queries, writing boilerplate copy—but even the best AI models were clearly not capable enough to precipitate widespread job displacement.

[Read: A chatbot is secretly doing my job]

Since then, however, two transformations have taken place. First, AI search became standard. Chatbots exploded in popularity because they could lucidly—though frequently inaccurately—answer human questions. Billions of people were already accustomed to asking questions and finding information online, making this an obvious use case for AI models that might otherwise have seemed like research projects: Now 300 million people use ChatGPT every week, and more than 1 billion use Google’s AI Overview, according to the companies. Further underscoring the products’ relevance, media companies—including The Atlanticsigned lucrative deals with OpenAI and others to add their content to AI search, bringing both legitimacy and some additional scrutiny to the technology. Hundreds of millions were habituated to AI, and at least some portion have found the technology helpful.

But although plain chatbots and AI search introduced a major cultural shift, their business prospects were always small potatoes for the tech giants. Compared with traditional search algorithms, AI algorithms are more expensive to run. And search is an old business model that generative AI could only enhance—perhaps resulting in a few more clicks on paid advertisements or producing a bit more user data for targeting future advertisements.

Refining and expanding generative AI to do more for the professional class—not just students scrambling on term papers—is where tech companies see the real financial opportunity. And they’ve been building toward seizing it. The second transformation that has led to this new phase of the AI era is simply that the technology, while still riddled with biases and inaccuracies, has legitimately improved. The slate of so-called reasoning models released in recent months, such as OpenAI’s o3-mini and xAI’s Grok 3, has impressed in particular. These AI products can be genuinely helpful, and their applications to advancing scientific research could prove lifesaving. Economists, doctors, coders, and other professionals are widely commenting on how these new models can expedite their work; a quarter of tech start-ups in this year’s cohort at the prestigious incubator Y Combinator said that 95 percent of their code was generated with AI. Major firms—McKinsey, Moderna, and Salesforce, to name just a handful—are now using it in basically every aspect of their businesses. And the models continue getting cheaper, and faster, to deploy.

[Read: The GPT era is already ending]

Tech executives, in turn, have grown blunt about their hopes that AI will become good enough to do a human’s work. In a Meta earnings call in late January, CEO Mark Zuckerberg said, “2025 will be the year when it becomes possible to build an AI engineering agent” that’s as skilled as “a good, mid-level engineer.” Dario Amodei, the CEO of Anthropic, recently said in a talk with the Council on Foreign Relations that AI will be “writing 90 percent of the code” just a few months from now—although still with human specifications, he noted. But he continued, “We will eventually reach the point where the AIs can do everything that humans can,” in every industry. (Amodei, it should be mentioned, is the ultimate techno-optimist; in October, he published a sprawling manifesto, titled “Machines of Loving Grace,” that posited AI development could lead to “the defeat of most diseases, the growth in biological and cognitive freedom, the lifting of billions of people out of poverty to share in the new technologies, a renaissance of liberal democracy and human rights.”) Altman has used similarly grand language recently, imagining countless virtual knowledge workers fanning out across industries.

These bright visions have dimmed considerably when put into practice: Elon Musk and the Department of Government Efficiency’s efforts to replace human civil servants with AI may be the clearest and most dramatic execution of this playbook yet, with massive job loss and little more than chaos to show for it so far. Meanwhile, all of generative-AI models’ issues with bias, inaccuracy, and poor citations remain, even as the technology has advanced. OpenAI’s image-generating technology still struggles at times to produce people with the right number of appendages. Salesforce is reportedly struggling to sell its AI agent, Agentforce, to customers because of issues with accuracy and concerns about the product’s high cost, among other things. Nevertheless, the corporation has pressed on with its pitch, much as other AI companies have continued to iterate on and promote products with known issues. (In a recent earnings call, Salesforce CEO Marc Benioff said the firm has “3,000 paying Agentforce customers who are experiencing unprecedented levels of productivity.”) In other words, flawed products won’t stop tech companies’ push to automate everything—the AI-saturated future will be imperfect at best, but it is coming anyway.

The industry’s motivations are clear: Google’s and Microsoft’s cloud businesses, for instance, grew rapidly in 2024, driven substantially by their AI offerings. Meta’s head of business AI, Clara Shih, recently told CNBC that the company expects “every business” to use AI agents, “the way that businesses today have websites and email addresses.” OpenAI is reportedly considering charging $20,000 a month for access to what it describes as Ph.D.-level research agents.

Google and Perplexity did not respond to a request for comment, and a Microsoft spokesperson declined to comment. An OpenAI spokesperson pointed me to an essay from September in which Altman wrote, “I have no fear that we’ll run out of things to do.” He could well be right; the Bureau of Labor Statistics projects AI to substantially increase the demand for computer and business occupations through 2033. A spokesperson for Anthropic referred me to the start-up’s initiative to study and prepare for AI’s effect on the labor market. The effort’s first research paper analyzed millions of conversations with Anthropic’s Claude model and found that it was used to “automate” human work in 43 percent of cases, such as identifying and fixing a software bug.

Tech companies are revealing, more clearly than ever, their vision for a post-work future. ChatGPT started the generative-AI boom not with an incredible business success, but with a psychological one. The chatbot was and is still possibly losing the company money, but it exposed internet users around the world to the first popular computer program that could hold an intelligent conversation on any subject. The advent of AI search may have performed a similar role, presenting limited opportunity for immediate profits but habituating—or perhaps inoculating—millions of people to bots that can think, write, and live for you.

Trump’s Attempts to Muzzle the Press Look Familiar

The Atlantic

www.theatlantic.com › ideas › archive › 2025 › 03 › trumps-press-freedom-hungary-orban › 682060

When Viktor Orbán gave a speech in 2022 at a Conservative Political Action Conference gathering in Budapest, he shared his secret to amassing power with Donald Trump’s fan base. “We must have our own media,” he told his audience.

As a Hungarian investigative journalist, I have had a firsthand view of how Orbán has built his own media universe while simultaneously placing a stranglehold on the independent press. As I watch from afar what’s happening to the free press in the United States during the first weeks of Trump’s second presidency—the verbal bullying, the legal harassment, the buckling by media owners in the face of threats—it all looks very familiar. The MAGA authorities have learned Orbán’s lessons well.

I saw the roots of Orbán’s media strategy when I first met him for an interview, in 2006. He was in the opposition then but had served as prime minister before and was fighting hard to get back in power. When we met in his office in a hulking century-old building that overlooked the Danube River in Budapest, he was very friendly, even charming. Like Trump, he is the kind of politician who knows how to connect with people when he thinks he has something to gain.

During the interview, his demeanor shifted. I still remember how his face went dark when I pushed on questions that he obviously did not want to answer. It was a tense exchange, but he reverted to his cordial mode when we finished the interview, and I turned off the recorder.

What happened afterwards was less friendly. In Hungary, journalists are expected to send edited interview transcripts to their interviewees. The idea is that if the interviewees think you took something they said out of context, they can ask for changes before publication. But in this case, Orbán’s press team sent back the text with some of his answers entirely deleted and rewritten. When my editors and I told them we wouldn’t accept this, they said they wouldn’t allow the interview to be published.

In the end, we published it without their edits. That was the last time I interviewed Viktor Orbán. And when he returned to power in 2010 after a landslide election victory, he made sure that he would never have to answer uncomfortable questions again.

One of the first pieces of legislation his party introduced was a media law that restructured how the sector is regulated in Hungary. The government set up a new oversight agency and appointed hard-line loyalists to its key positions. This agency later blocked proposed mergers and acquisitions by independent media companies, while issuing friendly rulings for pro-government businesses.

The Orbán government also transformed public broadcasting—which had previously carried news programs challenging politicians from all parties—into a mouthpiece of the state. The service’s newly appointed leaders got rid of principled journalists and replaced them with governing-party sympathizers who could be counted on to toe the line.

Then the government went after private media companies. Origo, a popular Hungarian news website, was one of its first targets. For many years, Origo—where I had been working when I conducted the 2006 Orbán interview—was a great place to do journalism. It was owned by a multinational telecommunications company and run by people who did not interfere with our work. If anything, they were supportive of our journalism. In 2009, after conducting some award-winning investigations, I was even invited to the CEO’s office for a friendly chat about the importance of accountability reporting.

But a few years after Orbán’s return to power, the environment changed. As we continued our aggressive—but fair—reporting, the telecommunications company behind Origo came under pressure from the government. Instead of sending encouraging messages, the outlet’s publisher started telling the editor in chief not to pursue certain stories that were uncomfortable for Orbán and his allies.

My colleagues in the newsroom and I pushed back. But after repeated clashes with the publisher over one of my investigations, into the expensive and mysterious travel of a powerful government official, the editor in chief was forced out of his job. I resigned, along with many fellow journalists, and soon the news site was sold to a company with close links to Orbán’s inner circle. Now Origo is unrecognizable. It has become the flagship news site of the pro-government propaganda machine, publishing articles praising Orbán and viciously attacking his critics.

Origo is part of an ecosystem that includes hundreds of newspapers and news sites, several television channels—including the public broadcasters and one of the two biggest commercial channels—and almost all radio stations. That’s not to mention the group of pro-government influencers whose social-media posts are distributed widely, thanks to financial resources also linked to the government.

This machine is not even pretending to do journalism in the traditional sense. It is not like Fox News, which still has some professional anchors and reporters alongside the openly pro-Trump media personalities who dominate the channel in prime time.

The machine built under Orbán has only one purpose, and it is to serve the interests of the government. There is hardly any autonomy. Editors and reporters get directions from the very top of the regime on what they can and cannot cover. If there is a message that must be delivered, the whole machine jumps into action: Hundreds of outlets will publish the same story with the same headline and same photos.

In 2022, Direkt36, the investigative-reporting center I co-founded after leaving Origo, wrote about one such example. In the story, which was reported by my colleague Zsuzsanna Wirth, we described an episode in which Bertalan Havasi, the prime minister’s press chief at the time, sent an email to the director of the national news agency.

​​“Hi, could you write an article about this, citing me as a source? Thanks!” Havasi wrote. (The instruction was about a relatively mundane matter: a letter that a European rabbi had sent to Orbán thanking him for his support.) Later, Havasi also told the agency what the headline and lead sentence should be. The news agency followed the instructions word for word.

A few years ago, I investigated the pro-government takeover of Index, another of Hungary’s most popular news sites. I obtained a recording in which the outlet’s editor in chief described to one of his employees how Index had received financial backing from a friend of Orbán’s, a former gas fitter who has become Hungary’s richest man thanks to lucrative state contracts. The editor in chief warned that Index had to be careful with news about Orbán’s friend because, without him, “there will be no one who will put money into” the outlet.

Just as Orbán explained in his CPAC speech, this sophisticated propaganda machine has played a crucial role in his ability to stay in power for more than 15 years. When the Organization for Security and Co-operation in Europe, a watchdog group of which the United States is a member, published its report on Hungary’s 2022 parliamentary elections, it pointed to the media as a major weakness in the country’s democratic system.

“The lack of impartial information in the media about the main contestants, the absence of debates among the major electoral competitors, and the independent media’s limited access to public information and activities of national and local government significantly limited voters’ opportunity to make an informed choice,” the election monitors concluded, after a vote that yet again cemented the power of Orbán’s ruling party.

What has happened in Hungary might not happen in the United States. Hungary, a former Eastern Bloc nation that broke free of oppressive Soviet control only three and a half decades ago, has never had such a robust and vibrant independent media scene as the one the U.S. has enjoyed for centuries. But if someone had told me when Orbán returned to power that we would end up with a propaganda machine where the free Hungarian media had once been, with many of the old outlets shut down or transformed into government mouthpieces, I would not have believed it.

And I see ominous signs in the U.S. that feel similar to the early phases of what we experienced here. When I read about the Associated Press being banned from White House events, that reminds me of how my colleagues at Direkt36 have been denied entry to Orbán’s rare press conferences. When I see the Washington Post owner Jeff Bezos cozying up to Trump, that reminds me of how big corporations and their wealthy executives, including the owner of my former workplace, bent the knee to Orbán.

When I read about ABC settling a Trump lawsuit of dubious merit—and CBS contemplating the same—it brings to mind the way the courts and the government itself can be used to manipulate and bully media organizations into submission.

Journalists and anyone else who cares about the free press must understand that democratic institutions are more fragile than they look, especially if they face pressure from ruthless and powerful political forces. This is particularly true for the news media, which is also being challenged by the technological revolution in how we communicate information. Just because an outlet has been around for decades and has a storied history does not mean that it will be around forever.

If any good news can be learned from Hungary’s unhappy experience, it is that unless your country turns into a fully authoritarian regime similar to China or Russia, there are still ways for independent journalism to survive. Even in Hungary, some outlets manage to operate independently from the government. Many of them, including the one I run, rely primarily on their audience for support in the form of donations or subscriptions. We learned that it is easy for billionaires and media CEOs to be champions of press freedom when the risks are low, but that you can’t count on them when things get tough. So we rely on our readers instead.

If they feel like what you are doing is valuable, they will be your real allies in confronting the suffocating power of autocracy.

Buy, Borrow, Die

The Atlantic

www.theatlantic.com › ideas › archive › 2025 › 03 › tax-loophole-buy-borrow-die › 682031

America’s superrich have always found ways to avoid paying taxes, but in recent years, they’ve discovered what might be the mother of all loopholes. It’s a three-step process called “Buy, Borrow, Die,” and it allows people to amass a huge fortune, spend as much of it as they want, and pass the rest—untaxed—on to their heirs. The technique is so cleverly designed that the standard wish list of progressive tax reforms would leave it completely intact.

Step one: buy. The average American derives most of their disposable income from the wages they earn working a job, but the superrich are different. They amass their fortune by buying and owning assets that appreciate. Elon Musk hasn’t taken a traditional salary as CEO of Tesla since 2019; Warren Buffett, the chair of Berkshire Hathaway, has famously kept his salary at $100,000 for more than 40 years. Their wealth consists almost entirely of stock in the companies they’ve built or invested in. The tax-law scholars Edward Fox and Zachary Liscow found that even when you exclude the 400 wealthiest individuals in America, the remaining members of the top 1 percent hold $23 trillion in assets.

Unlike wages, which are taxed the moment they are earned, assets are taxed only at the moment they are sold—or, in tax terms, “realized.” The justification for this approach is that unrealized assets exist only on paper; you can’t pay for a private jet or buy a company with stocks, even if they have appreciated by billions of dollars. In theory, the rich will eventually need to sell their assets for cash, at which point they will pay taxes on their increase in wealth.

That theory would be much closer to reality if not for step two: borrow. Instead of selling their assets to make major purchases, the superrich can use them as collateral to secure loans, which, because they must eventually be repaid, are also not considered taxable income. Larry Ellison, a co-founder of Oracle and America’s fourth-richest person, has pledged more than $30 billion of his company’s stock as collateral in order to fund his lavish lifestyle, which includes building a $270 million yacht, buying a $300 million island, and purchasing an $80 million mansion. A Forbes analysis found that, as of April 2022, Musk had pledged Tesla shares worth more than $94 billion, which “serve as an evergreen credit facility, giving Musk access to cash when he needs it.”

This strategy isn’t as common among the merely very rich, who may not have the expensive tastes that Ellison and Musk do, but it isn’t rare either. Liscow and Fox calculated that the top 1 percent of wealth-holders, excluding the richest 400 Americans, borrowed more than $1 trillion in 2022. And the approach appears to be gaining momentum. Last year, The Economist reported that, at Morgan Stanley and Bank of America alone, the value of “securities-backed loans” increased from $80 billion in 2018 to almost $150 billion in 2022. “The real question is: Why would you not borrow hundreds of millions, even billions, to fund the lifestyle you want to live?” Tom Anderson, a wealth-management consultant and former banker who specializes in these loans, told me. “This is such an easy tool to use. And the tax benefits are massive.”

[Annie Lowrey: Trump says his tax plan won’t benefit the rich—he’s exactly wrong]

You might think this couldn’t possibly go on forever. Eventually, the rich will need to sell off some of their assets to pay back the loan. That brings us to step three: die. According to a provision of the tax code known as “stepped-up basis”—or, more evocatively, the “angel of death” loophole—when an individual dies, the value that their assets gained during their lifetime becomes immune to taxation. Those assets can then be sold by the billionaire’s heirs to pay off any outstanding loans without them having to worry about taxes.

The justification for the stepped-up-basis rule is that the United States already levies a 40 percent inheritance tax on fortunes larger than $14 million, and it would be unfair to tax assets twice. In practice, however, a seemingly infinite number of loopholes allow the rich to avoid paying this tax, many of which involve placing assets in byzantine legal trusts that enable them to be passed seamlessly from one generation to the next. “Only morons pay the estate tax,” Gary Cohn, a former Goldman Sachs executive and the then–chief economic adviser to Donald Trump, memorably remarked in 2017.

“All of this is completely, perfectly legal,” Edward McCaffery, the scholar who coined the term Buy, Borrow, Die, told me. But, he said, the strategy “has basically killed the entire concept of an income tax for the wealthiest individuals.” The tax economist Daniel Reck, who has spent his career documenting the various ways the rich evade taxation, told me that Buy, Borrow, Die is “the most important tax-avoidance strategy today.” The result is a two-tiered tax system: one for the many, who earn their income through wages and pay taxes, and another for the few, who accumulate wealth through paper assets and largely do not pay taxes.

Much of the debate around American tax policy focuses on the income-tax rate paid by the very wealthiest Americans. But the bulk of those people’s fortunes doesn’t qualify as income in the first place. A 2021 ProPublica investigation of the private tax records of America’s 25 richest individuals found that they collectively paid an effective tax rate of just 3.4 percent on their total wealth gain from 2014 to 2018. Musk paid 3.3 percent, Jeff Bezos 1 percent, and Buffett—who has famously argued for imposing higher income-tax rates on the superrich—just 0.1 percent.

The same dynamic exists, in slightly less egregious form, further down the wealth distribution. A 2021 White House study found that the 400 richest American households paid an effective tax rate of 8.2 percent on their total wealth gains from 2010 to 2018. Liscow and Fox found that, excluding the top 400, the rest of the 0.1 percent richest individuals paid an effective rate of 12 percent from 2004 to 2022. (Twelve percent is the income-tax rate paid by individuals who make $11,601 to $47,150 a year.)

One solution to this basic unfairness would be to tax unrealized assets. In 2022, the Biden administration proposed a “billionaire minimum tax” that would have placed a new annual levy of up to 20 percent on the appreciation of even unsold assets for households with more than $100 million in wealth. Experts have vehemently debated the substantive merits of such a policy; the real problem, however, is political. According to a survey conducted by Liscow and Fox, most Americans oppose a tax on unrealized gains even when applied only to the richest individuals. The Joe Biden proposal, perhaps unsurprisingly, went nowhere in Congress. Making matters more complicated, even if such a policy did pass, the Supreme Court would very likely rule it unconstitutional.

[James Kwak: The tax code for the ultra-rich vs. the one for everyone else]

A second idea would be to address the “borrow” step. Last year, Liscow and Fox published a proposal to tax the borrowing of households worth more than $100 million, which they estimated would raise about $10 billion a year. The limitation of that solution, as the authors acknowledge, is that it would not address the larger pool of rich Americans who don’t borrow heavily against their assets but do take advantage of stepped-up basis.

That leaves the “die” step. Tax experts from across the political spectrum generally support eliminating the “stepped-up basis” rule, allowing unrealized assets to be taxed at death. This would be far more politically palatable than the dead-on-arrival billionaire’s minimum tax: In the same survey in which respondents overwhelmingly opposed broad taxes on unrealized assets during life, Liscow and Fox also found that nearly two-thirds of them supported taxing unrealized assets at death.

Even a change this widely supported, however, would run up against the iron law of democratic politics: Policies with concentrated benefits and distributed costs are very hard to overturn. That’s especially true when the benefits just so happen to be concentrated among the richest, most powerful people in the country. In fact, the Biden administration did propose eliminating stepped-up basis as part of its Build Back Better legislation. The move prompted an intense backlash from special-interest groups and their allied politicians, with opponents portraying the provision as an assault on rural America that would destroy family farms and businesses. These claims were completely unfounded—the bill had specific exemptions for family businesses and applied only to assets greater than $2.5 million—but the effort succeeded at riling up enough Democratic opposition to kill the idea.

The one guarantee of any tax regime is that, eventually, the rich and powerful will learn how to game it. In theory, a democratic system, operating on behalf of the majority, should be able to respond by making adjustments that force the rich to pay their fair share. But in a world where money readily translates to political power, voice, and influence, the superrich have virtually endless resources at their disposal to make sure that doesn’t happen. To make society more equal, you need to tax the rich. But to tax the rich, it helps for society to be more equal.