In 2019, Sam Altman said AI would “probably replace most of the jobs people do today.” In 2025, he said whole job categories would be “totally, totally gone.” On May 26 of this year, he reversed that view.
Speaking by video at a Commonwealth Bank of Australia conference, the OpenAI chief said: “I don’t think we’re going to have the kind of jobs apocalypse that some of the companies in our space advocate or talk about.” He said he had expected more impact on entry-level white-collar jobs by now, and that he was “delighted to be wrong.”
The shift is notable because it lines up with what two of the biggest research firms are now reporting. Gartner and Forrester disagree on some details, but both describe companies that cut staff for AI finding the hoped-for savings harder to realize than expected.
What Altman Said, and What Changed
The May remarks were widely reported, including by Reuters and Time magazine. In July, Altman wrote on X that he was “pretty sure” AI has so far created more jobs than it has eliminated, according to The Decoder.
He did not say disruption was over. He told the audience that “the jobs picture is likely to be very different than we thought,” and that many roles still depend on human contact. He has also criticized companies that blame AI for cuts he thinks have other causes (see “AI Takes the Blame”).
Why the change? Nobody outside OpenAI knows. Time magazine listed possible factors without confirming any: studies finding little AI effect on the labor market so far, mounting costs of running AI, and a public mood that has turned against the technology. One analyst quoted by Time magazine said it was unclear whether forecasts changed or only the narrative.
That caution is worth keeping in mind. OpenAI sells AI to employers and is raising capital. A softer position on job losses might help its position. The question is whether independent research supports there being fewer AI-related job losses than expected.
What Gartner Found
Gartner has published several findings about the impact of AI on the workplace this fall. Among the findings:
- Value is rare. In a survey of 2,002 CIOs and technology executives fielded May to July, only 13% said their enterprise has realized significant or transformational value from AI. Use-case success remains about one in five. Alicia Mullery of Gartner put it this way: “Volume is not value.”
- Stopping rules separate the leaders. CIOs with guidelines for when a pilot should not move to production were 2.3 times more likely to be among Gartner’s high performers. Only 31% have them. Guidelines for retiring low-value AI already in production raised the odds 2.6 times, and only 17% have those.
- Cost is rising, not falling. 58% of CIOs report more pressure to deliver AI cost improvements, and 51% expect AI to raise the total cost of ownership of technology.
- CEOs cannot show the return. In a survey of 203 CEOs and senior executives, 44% said they struggle to demonstrate AI financial returns. 63% pointed to radically redesigning the operating model as the path to more, and 52% to deploying AI agents as employees.
- Layoffs do not yield ROI. A Gartner study of 350 companies with revenue above $1 billion found about 80% had cut staff through automation efforts, and cutters were no more likely to report strong returns. Analyst Helen Poitevin said reductions “may create budget room, but they do not create return.” The best results came from making workers more productive, which Gartner calls people amplification.
- Entry-level hiring is shrinking. In a survey of 110 heads of HR, 22% said that at least one business leader in their company had stopped hiring entry-level workers because of AI automation. Gartner’s headline describes this as nearly a quarter of organizations, but the survey counts HR heads who report at least one such leader. It does not measure how many companies have cut entry-level hiring overall.
- Rehiring is coming. Gartner predicts that by 2029, 30% of employees who were laid off because AI replaced them will be rehired, often at much higher cost. It also predicts that by 2027, 75% of organizations that treat AI productivity gains as cost savings will be outpaced by competitors that reinvest the gains.
Gartner has also challenged the layoff narrative. It said fewer than 10% of layoffs were attributed to AI, and only 1% were tied to roles that actually disappeared (see “AI Takes the Blame”). Economic conditions, market pressures, and traditional cost cutting, it said, remain the primary drivers.
A related figure appeared in CIO Dive, which reported that AI-productivity-driven cuts were less than 1% of all 2025 layoffs, based on a Gartner review of more than one million layoffs. Gartner’s Tori Paulman said some AI-attributed cuts might be “AI-washing.” (See “Legit AI or AI Washing?” by Michael Neubarth.)
What Forrester Found
Forrester’s workforce claims come from its Predictions 2026 research on the future of work, plus two newer items on technology spending and data.
- Half of AI layoffs reversed. Forrester expects half of AI-attributed layoffs to be quietly reversed. In its words, much of that work will go to “lower-wage human workers, offshore or at lower salary.” It also warns that HR teams may face staffing cuts of up to 50%.
- Employer regret. Secondary coverage, including The Register and People Matters, reports that 55% of employers who cut staff citing AI now regret it, attributing the figure to Forrester.
- AI eats the budget. In an October 21 release, Forrester predicts AI will account for 20% of all business technology spend in 2027, in a market it puts at $6.1 trillion. A separate release says rising AI costs will force B2C leaders to scale back a third of their AI initiatives.
- Data is the bottleneck. A figure you may see attributed to Forrester says 85% of leaders believe fragmented data and systems must be unified for workplace AI to succeed, and 45% cite missing organizational context as the main reason for poor results. That comes from a survey of 310 IT decision-makers, conducted by Forrester Consulting for the intranet vendor Simpplr. It is vendor-commissioned research, and the sponsor sells what it recommends.
Where the Analysts Agree, and Where They Don’t
Both Gartner and Forrester say replacing people is more expensive and less reliable than the premise suggests, and both expect reversals. Both say the fix lies in redesigning work around people, not removing them.
They differ on the cost. Gartner expects rehired workers to cost more, at higher salaries or consultant rates. Forrester expects the work to return at lower pay or offshore. It will likely be a mix of both scenarios, with scarce skills commanding a premium and routine work moving elsewhere.
Forrester’s reversal forecast and Gartner’s 2029 rehiring figure are projections, and no one has yet counted how many jobs have actually come back. Other surveys point the same way but come with their own limits: A 600-person HR survey by Careerminds found about 36% of employers had rehired more than half of roles cut for AI, per People Matters.
Gartner’s 30% figure has also drawn skepticism. Computerworld reports that Paulman called it an average across organizations of all sizes, and said the figure for large enterprises could be roughly 40%. Other analysts questioned it. Valence Howden of Info-Tech said, “I’m not sure we can substantiate those numbers, since it’s much more of a guesswork statement than anything else.” Some noted that the same math implies 70% of AI-attributed cuts stay in place.
What It Means for Companies
- Do not book layoff savings before the work is redesigned. Gartner’s own data shows cutters do no better on ROI. Our AI ROI Measurement Playbook covers counting review time and total cost.
- Set stopping rules now. Gartner’s leaders define when a pilot should not graduate and when live AI should be retired. Most organizations have neither.
- Protect the early-career pipeline. If AI absorbs entry-level tasks, decide how new hires will learn the work. Our AI Adoption and Change Management Playbook covers role redesign.
- Fix the data before the agents. Weak access to organizational context is the most consistent cause of poor results across the vendor-commissioned research.
- Keep a record of AI-attributed cuts and the reasons. Regulators and employees are watching. Our AI Compliance Readiness Playbook shows what to document.
Bottom Line
The analyst findings point to the conclusion that AI has not produced the predicted job collapse so far, and that the companies that treated it as a headcount tool are not reporting better returns.
Gartner still expects entry-level hiring to erode and roles to broaden, while Forrester expects some work to come back on worse terms for workers. Altman, whose earlier forecasts shaped the debate, now says his predictions were off.
Our view is that AI requires human guidance and oversight. Realizing the optimum benefits of AI requires skilled application of AI automation, as well as talented strategists, writers, and editors.
WorkplaceAI continues to report on the state of AI in the workplace, including the original expectations of AI and the disappointing real-world experiences of organizations so far (see the recent report showing how “Executives Fear for Their Jobs Over AI.”)
Sources: AI Pulse · Big Picture · workplaceai.ai. Altman: Time magazine (May 26, 2026); IT Pro (June 1, 2026); BNN Bloomberg and Euronews on the Reuters report (May 26, 2026); The Decoder (July 12, 2026). Gartner: CIO survey release via Webull/Barchart (Oct. 19, 2026); CEO survey release (Oct. 20, 2026); Future of Work release (Sept. 9, 2026) and CIO Dive (Sept. 16, 2026); entry-level hiring release (July 27, 2026) and People Matters; layoffs and ROI study via The Register (May 6, 2026) and Fortune (May 11, 2026). Forrester: Predictions 2026 future of work blog and The Register (Oct. 29, 2025); People Matters (July 31, 2026); The Next Web (Aug. 23, 2026); 2027 technology and security predictions release (Oct. 21, 2026); Simpplr press release on the Forrester Consulting study (2026). Gartner statement on layoffs (early Oct. 2026); Computerworld on Gartner’s rehiring forecast.