> ## Content Index > Fetch the complete content index at: https://www.bushletter.com/llms.txt > Use this file to discover other available public pages before exploring further. # OpenAI boss admits he was wrong about how fast AI would take jobs - URL: https://www.bushletter.com/openai-boss-admits-he-was-wrong-about-how-fast-ai-would-take-jobs/ - Published: 2026-08-25T05:00:00.000Z - Updated: 2026-08-25T04:59:59.000Z - Description: When GPT-4 launched in 2023, Sam Altman expected a rapid, visible restructuring of white-collar work. Businesses would retool, roles would vanish, whole sectors would be remade faster than regulators or HR departments could track. - Author: Editor - Tags: Technology, US, Sam Altman ![Alex Mercer](https://res.cloudinary.com/dz77sb7j1/image/upload/v1774262566/bushletter/authors/alex-mercer.png) By **Alex Mercer** ยท 2026-08-24 TLDR OpenAI chief Sam Altman says he was too ambitious about how fast AI would reshape the economy after GPT-4, citing deep organisational inertia as the main brake. Goldman Sachs estimates AI still cut roughly 16,000 US jobs a month over the past year, even as the predicted white-collar apocalypse has not arrived. KEY TAKEAWAYS 01Altman publicly admitted AI economic disruption is unfolding far more slowly than he expected after GPT-4. 02His exact words: 'We've all been too ambitious on timelines' due to the economy's deep inertia. 03At a Commonwealth Bank event in May 2026, Altman said he was 'delighted to be wrong' about white-collar job losses. 04Goldman Sachs Research estimated AI removed roughly 16,000 US jobs a month over the past year. 05Slow adoption and measured job losses are compatible: inertia dampens speed, but displacement is still occurring. ## The forecast that did not arrive When GPT-4 launched in 2023, Sam Altman expected a rapid, visible restructuring of white-collar work. Businesses would retool, roles would vanish, whole sectors would be remade faster than regulators or HR departments could track. That forecast has not played out on schedule, and Altman is now saying so plainly. Speaking on 23 August 2026, Altman said society and the economy will adapt more slowly than he anticipated, framing economic inertia as the primary constraint.[\[1\]](https://www.usetranscribe.io/yt/kG8AoExkX40/sam-altman-openai-ai?ref=bushletter.com) "One of them is the economy just has so much inertia," Altman said, before adding: "I think AI is one of the most incredible technologies humanity has ever invented. Society and the economy will adapt more slowly."[\[1\]](https://www.usetranscribe.io/yt/kG8AoExkX40/sam-altman-openai-ai?ref=bushletter.com) ## A pattern of revision Altman has walked back his economic timeline before. At Commonwealth Bank's Accelerate AI event in May 2026, he offered an unusually candid self-assessment: "My scorecard, at the highest level would be we've been roughly right on technological predictions and pretty wrong on the social and economic implications."[\[2\]](https://www.commbank.com.au/articles/newsroom/2026/05/sam-altman-close-ai-gap.html?ref=bushletter.com) That remark drew attention because the directional error runs counter to the dominant Silicon Valley narrative. Technologists routinely underestimate how fast technology spreads; Altman is conceding the opposite failure. OpenAI's read on the capability curve has held up. Its read on human and institutional behaviour has not. ## Relief and responsibility One specific prediction Altman got wrong concerned entry-level white-collar jobs. Early commentary around large language models suggested coding assistants, legal research tools and content-generation software would quickly hollow out junior professional roles. At the Commonwealth Bank event, Altman said he is "delighted to be wrong about that," referring directly to those white-collar displacement forecasts.[\[2\]](https://www.commbank.com.au/articles/newsroom/2026/05/sam-altman-close-ai-gap.html?ref=bushletter.com) The admission carries weight because Altman sits at the centre of the system producing these tools. Whether the delay is permanent or simply a lag before a steeper adjustment curve remains an open question. ## What Goldman Sachs found in the data Altman's revised optimism does not mean AI is leaving employment untouched. Goldman Sachs Research estimated that AI reduced US monthly payroll growth by roughly 16,000 jobs over the past year, based on its analysis of labour market data through April 2026.[\[3\]](https://www.goldmansachs.com/insights/articles/the-jobs-ai-is-likely-to-boost-and-those-it-may-disrupt.html?ref=bushletter.com) Sixteen thousand jobs a month is not a trivial figure. Annualised, it approaches 200,000 positions. Yet in an economy that adds hundreds of thousands of jobs in strong months, that signal sits within the noise of ordinary churn, and Goldman Sachs's finding is consistent with Altman's revised timeline rather than contradicted by it. A slow, steady drag on payroll growth is precisely what inertia-constrained adoption would produce. Goldman Sachs also found AI is likely to boost employment in some categories while disrupting others, a distribution that further mutes aggregate headline numbers.[\[3\]](https://www.goldmansachs.com/insights/articles/the-jobs-ai-is-likely-to-boost-and-those-it-may-disrupt.html?ref=bushletter.com) Roles requiring physical presence, complex judgment under uncertainty, or ongoing client relationships appear more insulated, while structured, repetitive knowledge work is bearing most of the early pressure. ## Inertia as a systems problem Altman's use of the word inertia is doing more analytical work than it might first appear. In engineering terms, inertia is not resistance born of ignorance; it is the property of a system with mass and momentum that requires sustained force to change direction. Large organisations have procurement cycles, compliance obligations, workforce contracts, software dependencies and risk management frameworks that all interact. That structural reality explains why even companies that have publicly committed to AI integration report patchy internal deployment. Pilots succeed, then stall at procurement. Legal sign-off slows rollout. Integration with legacy data systems proves harder than demos suggested. The technology is ready; the surrounding systems are not. Altman's concession on timelines is, read this way, less a retreat than a more accurate systems diagnosis. Altman said "we've all been too ambitious on timelines," and the "we" in that sentence includes not just OpenAI but the full ecosystem of consultants, investors and CIOs who co-authored the disruption narrative.[\[1\]](https://www.usetranscribe.io/yt/kG8AoExkX40/sam-altman-openai-ai?ref=bushletter.com) ## What the revised picture looks like Read together, Altman's words and the Goldman Sachs data point to a displacement curve that is real but shallow in slope. AI is reforming work in measurable ways: tasks are shifting, role descriptions are changing, and some positions are not being replaced when vacated. The cliff-edge scenario, in which a GPT-4 equivalent triggers a visible quarterly spike in white-collar unemployment, has not materialised, and Altman is no longer predicting it will. For Australian businesses watching the US experience as a lead indicator, the practical implication is time. The technology is available now, but the organisational, regulatory and cultural infrastructure required to deploy it at scale is still being built, and nothing in Altman's 23 August remarks suggests that gap will close quickly. SOURCES & CITATIONS 1. [Sam Altman on AI timelines and economic inertia (transcript)](https://www.usetranscribe.io/yt/kG8AoExkX40/sam-altman-openai-ai?ref=bushletter.com) 2. [Sam Altman at CommBank Accelerate AI event, May 2026](https://www.commbank.com.au/articles/newsroom/2026/05/sam-altman-close-ai-gap.html?ref=bushletter.com) 3. [Goldman Sachs: The jobs AI is likely to boost and those it may disrupt](https://www.goldmansachs.com/insights/articles/the-jobs-ai-is-likely-to-boost-and-those-it-may-disrupt.html?ref=bushletter.com) FREQUENTLY ASKED QUESTIONS What exactly did Sam Altman admit about AI and jobs? Altman said he and others in the industry were "too ambitious on timelines" when predicting how quickly AI would disrupt the economy after GPT-4\. He credited the economy's deep inertia as the main reason adoption has been slower than forecast. Has AI caused any job losses, even if the 'apocalypse' has not arrived? Yes. Goldman Sachs Research estimated AI reduced US monthly payroll growth by roughly 16,000 jobs over the past year. That is a real effect, though it sits within the normal range of economic churn rather than producing a visible employment spike. Why does organisational inertia slow AI adoption? Large organisations face procurement cycles, compliance obligations, legacy system integration and workforce contracts that all interact. Even when AI tools are technically ready, the surrounding institutional infrastructure takes considerably longer to adapt. What did Altman say at the Commonwealth Bank event in May 2026? At Commonwealth Bank's Accelerate AI event, Altman said his team was "roughly right on technological predictions and pretty wrong on the social and economic implications." He also said he was "delighted to be wrong" about his earlier predictions on entry-level white-collar job losses. ![Alex Mercer](https://res.cloudinary.com/dz77sb7j1/image/upload/v1774262566/bushletter/authors/alex-mercer.png) [Alex Mercer](https://bushletter.com/author/alex-mercer/?ref=bushletter.com) Alex Mercer writes about technology, energy and infrastructure. He likes the physical end of the story: the plants, the grids and the machines that everything else depends on.