Two AI Giants, Two U-Turns: What's Really Going On?
Not long ago, the leaders of two of the world's most powerful artificial intelligence companies were sounding the alarm. Sam Altman of OpenAI and Dario Amodei of Anthropic both warned, publicly and emphatically, that AI was on a collision course with white-collar employment. Millions of jobs, they suggested, could be wiped out — particularly at the entry level. Those predictions made headlines, sparked congressional debates, and fueled widespread anxiety among workers across industries.
Now, both executives are telling a very different story. And the timing — with OpenAI and Anthropic both reportedly preparing for major initial public offerings — has not gone unnoticed.
Sam Altman Admits He Was "Pretty Wrong"
Speaking at a Commonwealth Bank of Australia event in Sydney on May 26, OpenAI CEO Sam Altman offered a candid admission that surprised many observers. "I thought there would have been more impact on entry-level white-collar jobs being eliminated by now than has actually happened," Altman said. He went further, adding, "I'm delighted to be wrong about this."
This is a notable shift from the tone Altman and OpenAI have struck in recent years. For much of the early AI boom, conversations around large language models were saturated with warnings about automation displacing knowledge workers — customer service representatives, junior analysts, paralegals, copywriters, and more. Altman himself was among those who acknowledged that AI would fundamentally reshape the workforce. His Sydney remarks suggest he now believes that disruption is arriving more slowly, or in a more manageable form, than previously expected.
Whether this is a genuine reassessment based on observed labor market data, or a strategic softening of messaging ahead of a public offering, is a question worth asking seriously.
Dario Amodei's Dramatic Reversal
Anthropic CEO Dario Amodei's evolution on this topic is even more striking. As recently as last year, Amodei warned that AI could eliminate up to half of all entry-level white-collar jobs and push unemployment rates into double digits — a genuinely alarming forecast that drew significant media attention and reinforced fears about AI's societal impact.
Today, Amodei is framing AI automation in an almost entirely different light. Rather than emphasizing job destruction, he now stresses productivity multiplication. In a recent interview reported by Fortune, Amodei offered what amounts to a reinterpretation of the displacement narrative. "If you automate 90% of the job, then everyone does the 10% of the job," he said. "And the 10% kind of expands to be 100% of what people do and kind of 10-times their productivity."
This reframing leans heavily on economic concepts like the Jevons Paradox — the idea that increased efficiency in resource use tends to increase overall consumption of that resource, rather than reducing it. Applied to labor, the argument is that AI will make workers so much more productive that demand for human work will actually grow, not shrink. It is a compelling theory, and one with historical precedent. But it is also one that conveniently positions AI as an unambiguous economic good — precisely the kind of story that plays well to public market investors.
Why the Timing Matters: IPOs on the Horizon
Both OpenAI and Anthropic are reportedly preparing IPOs, with valuations expected to reach staggering figures. OpenAI has been valued at approximately $300 billion in private markets, while Anthropic's valuation has climbed sharply following major investment rounds from Amazon and Google.
For companies seeking to attract institutional investors and retail shareholders, the narrative surrounding AI's societal impact matters enormously. A company whose CEO has predicted mass unemployment and double-digit jobless rates faces uncomfortable questions — about regulation, about public backlash, about the long-term sustainability of its business model in a society destabilized by its own products.
By contrast, a company whose CEO emphasizes productivity gains, human augmentation, and economic growth is telling a story that investors, policymakers, and the general public are far more receptive to. It is a story that invites optimism rather than fear, and optimism is far easier to price into an IPO prospectus.
This does not necessarily mean either Altman or Amodei is being dishonest. Minds can change. New data can alter forecasts. But the confluence of dramatically softened messaging with imminent public offerings is a pattern that warrants scrutiny.
What the Labor Market Actually Shows
The honest answer is that AI's impact on employment so far has been uneven and genuinely difficult to measure. Certain roles — particularly in content creation, basic coding, and data entry — have seen reduced hiring even as overall white-collar employment has remained relatively stable. Some companies have openly cited AI tools as a reason for leaner headcounts. Others have expanded their workforces while deploying AI extensively.
Economists broadly agree that technological disruption tends to create new categories of work over time, even as it eliminates existing ones. The industrial revolution, the advent of computing, and the rise of the internet all followed this pattern. The debate is not whether AI will change work — it clearly already is — but how fast, how unevenly, and who bears the cost of the transition.
The Risk of Premature Reassurance
There is a meaningful danger in AI executives publicly downplaying job displacement concerns at precisely the moment those concerns are most acute. Workers and policymakers need accurate, credible information to prepare for what is coming. Retraining programs, social safety nets, and regulatory frameworks all take time to build. If the dominant message from AI's most influential voices shifts prematurely to "everything will be fine," the institutions and policies needed to soften the disruption may arrive too late for those most affected.
It is worth noting that Amodei's earlier warnings — the ones he is now walking back — may have been closer to the truth than his current framing. Several economists and labor researchers have cautioned that productivity gains from AI, while real, do not automatically translate into broadly shared prosperity. They can just as easily concentrate wealth further upward while leaving displaced workers behind.
What Workers and HR Leaders Should Take Away
For HR professionals, workforce planners, and individual workers trying to make sense of these shifting signals, a few principles are worth holding onto. First, treat executive messaging around AI with appropriate skepticism, particularly when financial incentives are clearly in play. Second, invest in adaptability — the specific shape of AI's workforce impact remains genuinely uncertain, and versatility is the best hedge. Third, pay attention to what companies are actually doing with their headcounts, not just what their CEOs are saying on stages in Sydney or in columns in Fortune magazine.
The AI revolution is real. Its labor market consequences are real. Whether those consequences are being accurately described by the people building the technology — especially as those people prepare to sell billions of dollars in stock to the public — is a question every worker, investor, and policymaker should be asking right now.