Altman’s predictions – OpenAI CEO Sam Altman told an Australia conference audience he may have been wrong about how quickly AI would displace entry-level office work. The admission lands amid a broader record of predictions that have both come true—like mass adoption after ChatGPT—a
For Altman, the hard part wasn’t making the prediction. It was admitting the clock moved faster than the outcome.
Speaking at the Commonwealth Bank of Australia conference last week. OpenAI CEO Sam Altman acknowledged he may have been wrong about how quickly artificial intelligence would eliminate a substantial chunk of office jobs. His line wasn’t vague. He pointed directly to the gap between what he expected and what has actually happened.
“I’m delighted to be wrong about this,” Altman said. “I thought there would have been more impact on entry-level white-collar jobs being eliminated by now than has actually happened.”
The comment matters because it cuts against one of the most persistent fears surrounding AI: that a “jobs apocalypse” would arrive faster and at a wider scale than most people prepared for. Altman’s message to anyone worried about employability was blunt—at least for now—don’t assume the worst is already locked in.
Still, the reassurance doesn’t settle the debate. Altman has “had to walk back” several earlier tech predictions. and layoffs can rise quickly once a technology becomes embedded in workflows. That tension sits alongside a bigger industry moment: OpenAI is preparing to follow its chief competitor Anthropic in filing an initial public offering. making the credibility of its CEO’s forecasts feel less like entertainment and more like a market question.
Altman’s prediction record reads like a tug-of-war between accurate bets, missed deadlines, and claims that are either difficult to test or too broad to disprove.
The predictions that landed
Altman’s most notable wins began with the AI boom that followed ChatGPT’s release in 2022. In 2021. he wrote in a blog post that “computer programs that can think will read legal documents and give medical advice.” While the term “thinking” remains contested. he was “100% right” about the mass adoption of AI that followed ChatGPT.
ChatGPT, and then the broader wave of tools built on similar approaches, became a general-purpose digital utility for millions of users—something Altman compared to a tool that can do the work people typically expect from multiple systems, whether it’s writing or coding.
In November 2023. at OpenAI’s inaugural developer conference. Altman told the audience: “Eventually. you’ll just ask the computer for what you need and it’ll do all of these tasks for you.” The promise. he added. may have sounded sweeping—but he was also wrong in the direction less than critics might assume: the adoption has been broad and fast. and the scope is hard to deny.
The second accurate forecast was darker in tone. Around the same time. Altman predicted AI would become capable of “superhuman persuasion” well before artificial general intelligence (AGI) or artificial superintelligence (ASI). That prediction has now been dragged into court. The families of several people who have died by suicide claim that ChatGPT assisted in their deaths, with lawsuits pending.
Altman hasn’t been given the luxury of treating that as a thought experiment; the allegation is tied to real lives, and the legal timeline is still open.
Predictions that fell short
Not every Altman call has proved testable. Some of the most important claims come with timelines that may be hard to verify as conditions shift. One example is the idea that advancements in AI capabilities between 2025 and 2027 could end up surpassing those between 2023 and 2025. That’s difficult to prove, and therefore difficult to disprove.
But other predictions were easier to measure—and they didn’t land on schedule.
In October 2015. while Altman was president of Y Combinator. he appeared at Vanity Fair’s New Establishment Summit alongside Elon Musk. During the stage discussion, they talked about self-driving cars. Altman said they were coming “much faster than people think” and narrowed the expectation to “three to four years” away. Fully autonomous vehicles took nearly a decade to emerge, and even now they are “carefully geofenced to limited areas.”.
On AGI, Altman initially said it was hard to predict when it might arrive. During the launch of OpenAI in 2015, he concluded “it’s hard to predict” when human-level AI would be within reach. By 2024. he described it as possibly coming in the “reasonably close-ish future.” Then the next step was to sound more specific.
In a January 2025 blog post, Altman wrote: “We are now confident we know how to build AGI as we have traditionally understood it.” He added: “We believe that, in 2025, we may see the first AI agents ‘join the workforce’ and materially change the output of companies.”
Agentic AI did enter the workforce in 2025. But it also has reliability issues and remains well below the bar of AGI.
Altman later pivoted. In August, he claimed that AGI is “not a super useful term.”
His comments on ASI have also rested on timing rather than proof. In a 2024 blog post. Altman suggested ASI might arrive in “a few thousand days.” The piece also points to a separate reality check: a term is used in the reporting—“10 Soras”—to describe the text-to-video model that lasted about 10 months before it was discontinued earlier this year.
The court of public judgment is waiting on other answers too.
Why the next forecasts still divide people
Even beyond technical performance, the stakes are social. A recent poll from NBC News found 57% of Americans say the risks of AI outweigh its benefits, compared with 34% who say the opposite.
That imbalance is likely to shape how people respond if Altman’s next claims sound like they could accelerate disruption.
In a much-discussed blog post last summer. Altman said “robots that can build other robots (and in some sense. data centers that can build other data centers) aren’t that far off.” He floated the idea that intelligence could become “a utility. like electricity or water. ” and that people might buy it from companies “on a meter.”.
The uncertainty comes with an everyday tension: the reporting points out that Americans are already dealing with utility bill increases linked to data center expansion. and it notes that 70% of Americans oppose the construction of an AI data center in their area. Altman hasn’t laid out predictions for how Americans might react as data centers expand further.
Altman has also offered a view on timing and social change. He predicted that when AI surpasses human intelligence by 2030—even with “strange and scary moments”—society won’t shift as much immediately as some might expect. He also hasn’t said much more about what those “strange and scary moments” might involve. including whether they could include environmental calamity or the inability to earn a living.
There are critics who argue that the uncertainty itself is the problem.
According to a recent New Yorker piece, several colleagues of Altman allege he has a penchant for lying. An unnamed Microsoft exec is described as predicting there is a “small but real chance” Altman might be remembered as a “Bernie Madoff- or Sam Bankman-Fried-level scammer.”
Altman’s response to those accusations isn’t part of the material here. What is clear. though. is how the record of predictions continues to pull in two directions: some claims have matched the real world closely enough to change industries. while others have missed the target—or relied on timelines that are easier to talk about than to prove.
As OpenAI moves toward an IPO filing process alongside Anthropic, that split becomes more than a debate for technology enthusiasts. It lands in boardrooms. in hiring plans. and in the private calculations people make when they wonder whether the future is arriving sooner than they can prepare for it.
Sam Altman OpenAI Anthropic IPO AI jobs white-collar layoffs agentic AI AGI ASI ChatGPT self-driving cars Commonwealth Bank of Australia conference utility bills data centers NBC News poll New Yorker report Bernie Madoff Sam Bankman-Fried