Copy article

Some Silicon Valley AI CEOs have turned down the volume on the job apocalypse rhetoric. Why and why now?

ended 09. May 2026

Silicon Valley's AI chiefs spent two years telling you their products would end your career. Now they'd like your investment instead. 

Here are two statements in the last week:

  • Sam Altman, OpenAi CEO, 1 May 2026, on X: "we want to build tools to augment and elevate people, not entities to replace them. I think a lot of people are going to be busier and hopefully more fulfilled than ever. And jobs doomerism is likely long-term wrong." 
  • Jensen Huang, Nvidia CEO, 2 May 2026 in Fortune: Jensen Huang, Nvidia CEO, 2 May 2026: "Scaring people into believing that the technology will pose an existential threat to humanity, destroy democracy or eliminate 50% of entry-level jobs is ridiculous." On his fellow CEOs: "They're made by people who are like me, CEOs, and somehow because they became CEOs you adopt a God complex, and before you know it you know everything."

Huang estimated AI has created more than half a million jobs in recent years and cited Indeed data showing demand for software engineers is increasing.

Altman went from "your work isn't real" to "you'll be more fulfilled than ever" in six months. Huang is openly calling the doom predictions ridiculous and accusing the people who made them of delusion, possibly aiming at Anthropic's Dario Amodei's 50% white-collar job loss prediction. 

Something has clearly shifted. The question beyond why is whether any of it, the doom or the optimism, was ever based on what the technology actually does. Given rising levels of anxiety about AI in the global general population, this positioning matters a great deal.

Bloomberg's Parmy Olson described the pattern in April: the doom rhetoric was "a paradoxical marketing strategy" where fear turned out to be "the ultimate sales pitch." That paradox makes more sense than it looks. If your product can automate everyone's job, it must be extraordinarily powerful. The terror is the value proposition. Every prediction of mass unemployment was also an advert. 

So why the U-turn now? 

Both OpenAI and Anthropic are preparing for potential IPOs. Anthropic has hired law firm Wilson Sonsini. OpenAI restructured to for-profit last year. And IPO preparation means Wall Street due diligence, people in suits asking hard questions about revenue, retention, and whether your public statements create material risk for a prospectus. You can tell venture capitalists your product will eat the world. You cannot tell the SEC the same thing and then sell shares to pension funds. 

There is speculation from Cal Newport Professor of Computer Science at Georgetown University that East Coast and Wall Street mentality has shown up and said OpenAi must operate like a real business or we don't underwrite you. He says he is “picking up more of these signals” in his industry research.

Another reason is people using AI in the field are not finding it to be the job satisfaction and productivity silver bullet the glossy marketing promises, and thus public opinion on the benefits of AI has followed the rhetoric off a cliff. The terror factor marketing has bitten Silicon Valley hard.

Marc Bara, another writer felt that the independent evidence suggests neither the doom nor the new optimism tracks reality. 

  • A randomised trial by nonprofit METR found experienced developers using AI tools were 19% slower, not faster, while believing they'd sped up by 20%. (METR later announced it was abandoning its experimental design because developers refused to work without AI, destroying the control group. Their revised estimate suggested a possible speedup but they described their own evidence as "only very weak.")
  • 55% of chief executives reported no measurable benefit from AI deployment (IT Pro analysis including PwC research).
  •  Gartner placed AI in the "Trough of Disillusionment" for all of 2026. 

If the doom claims weren't based on operational evidence, and the new optimism isn't either, both positions look like marketing calibrated to whoever's writing the next cheque.

There's also a question about where the doom language came from in the first place. Professor Newport has a hypothesis from his personal observation of the space that it didn't start as a sales pitch. It originated in the effective altruism and existential risk communities, who undertook serious, if speculative, academic work on catastrophic threats to humanity. 

The same intellectual tradition that modelled asteroid impacts and pandemic preparedness turned its attention to artificial general intelligence. 

Several AI CEOs came directly from those circles or were heavily influenced by them. The language of civilisational risk, existential threat, and superintelligence that sounds so alarming in a product launch started life as philosophical caution about hypothetical future systems, the kind of thing you'd discuss at a conference on long-term species survival much like a debating circle topic. 

Newport suggests it got picked up, stripped of its caveats, and welded onto marketing for autocomplete tools that still hallucinate court cases. The sci-fi framing lent gravity and inevitability to products that don't yet reliably summarise a PDF. It worked commercially until the public believed it too literally, turned hostile, and the IPO bankers could well have said to put an end to AI doomerism.

We'd like your views:

  • Altman called displaced white-collar work "not real work" in October. He now says people will be "more fulfilled than ever." When a CEO of a company preparing to go public contradicts himself this rapidly, is it an evolving opinion or a material governance risk?
  • Huang says the doom predictions are "ridiculous" and driven by a "God complex." If he's right, what does that say about every hiring freeze, career pivot, and government policy drafted in response to those predictions over the past two years?
  • Independent research shows AI making some experienced workers slower, not faster, and a majority of CEOs reporting no measurable benefit. Were the doom predictions ever based on what the technology does, or on what companies needed investors to believe?
  • The doom narrative borrowed its language from legitimate existential risk research about hypothetical superintelligence. Now that the CEOs who popularised it are walking it back for commercial reasons, has genuine AI potential for good been damaged by association in the court of public opinion?
  • If you changed your business strategy based on CEO predictions of mass automation, what would it take for you to trust their new, optimistic version of the same product, with many more humans in the loop?

5 responses from the Newspage community

Copy all

Star Quote
Copy

Colette Mason
AI Ethics Consultant at Clever Clogs AI
The real casualty of this rhetoric circus isn't CEO credibility, it's the AI that actually works ntarred with the same brush. Every month a hospital trust delays an AI diagnostic tool because the public mood is hostile, or a farming cooperative won't touch precision agriculture because "AI" now means job loss, someone pays for that delay with their health, their crop yield, or their independence. AI-powered screen readers, real-time transcription, early cancer detection. None of these got a look-in while the CEOs were competing to see who could sound most apocalyptic.

The doom marketing didn't just inflate valuations. It trained an entire population to flinch at the word "AI" before asking what it does. Rebuilding that trust will take years and cost the very people these tools were built to help. The CEOs get to revise their positioning with a tweet. The stroke patient whose AI-assisted rehab programme got defunded because the board got nervous doesn't get a revised prognosis.
Copy

Editor
Katrina Young
Futurist | Chief Technology Officer CTO/CDO Ethical AI & Digital Transformation Strategist at KYC Digital
The shift from “AI will replace jobs” to “AI will augment people” reflects how quickly the market’s understanding of AI deployment is evolving. What matters now is not whether optimism or caution is right, but whether organisations are making workforce and operational decisions based on validated evidence rather than broad narratives about what AI may eventually become. Many businesses are discovering the gap between demonstration and deployment. Generating outputs is one thing. Building secure, compliant and reliable systems that function inside real organisations is another. Governance, cyber security, QA, human oversight and accountability still matter. Independent studies are already showing mixed productivity outcomes in some areas, suggesting AI’s impact may be far more uneven and role-specific than either the most optimistic or pessimistic predictions implied.
Copy

Editor
Mitali Deypurkaystha
AI Governance Strategist & CMO at AISensa
Nearly 90% of firms say AI has had no measurable impact on productivity or jobs over the past three years, according to a 2026 Fortune executive study. Yet Silicon Valley spent that same period selling a white-collar wipeout with absolute certainty. As an AI strategist, one of the biggest surprises was discovering how much of my job became calming frightened staff and reassuring business owners that productivity gains did not mean mass redundancies. People genuinely believed the apocalyptic rhetoric coming from tech’s billionaire class. Now the same CEOs who warned of economic extinction are suddenly talking about “fulfilment” and “augmentation”. It leaves ordinary workers feeling like pawns in a giant commercial narrative that keeps changing whenever financial incentives do. It resembles a kind of corporate, Orwellian Ministry of Truth, where yesterday’s certainty quietly disappears, and the public is expected to forget they were ever told their careers were doomed in the first place.
Copy

Premium
Debbie Porter
Managing Director at Destination Digital Marketing
I began my 'internet' career in the early 90s at the Daily Mail building websites, when websites weren't really a thing. It wasn't long before journalists became hostile to the web team who were putting out news coming in on the wires as soon as it broke rather than waiting for the publication deadlines. The Evening Standard and it's 4pm edition to catch homebound commuters was the first casualty when it sold to the Express for 1p. There was no longer an audience for the physical newspaper that carried news everyone had already read on their lunch break.

The times were changing, and this AI seachange is just the same as the dot.com revolution 30 years ago. Where one traditional process died off, a new digital age took over and three decades later we're all still standing - we all still have jobs. They just look a little different to how we imagined they would back then. Marketing of AI needs to move from scaremongering to become a friend-not-foe and this change of tack reflects that.
Copy

Rohit Parmar-Mistry
Founder at Pattrn Data
Silicon Valley has not discovered humility. It has discovered IPO risk. For two years, job apocalypse rhetoric made AI sound inevitable, terrifying and therefore valuable. Now the same CEOs need pension funds and public markets to believe these systems are useful infrastructure, not a social grenade on subscription.

The truth is that both stories were marketing. The doom story exaggerated capability to sell power. The new optimism softens the same pitch to sell safety and investability. Neither tells us much about what AI actually does inside a business, where most failures are duller: bad data, weak processes, no accountability, and managers pretending a chatbot is a strategy.

In our AI Audits, the question is never whether AI will replace everyone. It is whether anyone can explain the workflow, the risks and the human sign-off. People do not fear AI because they are backward. They fear being experimented on by companies that change the story whenever the funding round changes.