OpenAI and Anthropic CEOs both admit their mistakes: AI has not eliminated white-collar jobs, and what they said before the IPO has changed

OpenAI CEO Sam Altman and Anthropic CEO Dario Amodei have successively retracted their previous predictions that AI will eliminate white-collar jobs. Altman said he was 'horribly wrong' and Amodei argued that AI is actually a workload multiplier rather than a replacement. The reversal of attitudes comes at a critical time for both companies as they prepare for their respective IPOs.

Core conclusion

Over the past year, two of the AI industry’s loudest voices—OpenAI’s Sam Altman and Anthropic’s Dario Amodei—have publicly warned that AI will eliminate large numbers of white-collar jobs. But in May 2026, the two admitted: They were wrong.

Key Points

Altman admits: "I thought white-collar entry-level jobs were going to be decimated, but that hasn't happened yet."
Amodei reverses position: AI does not eliminate jobs, but increases everyone’s output by 10 times, and the job content itself will expand
Goldman Sachs CEO David Solomon has long insisted that AI will not cause mass unemployment, but now he seems even more prescient
The change in attitude of the two CEOs happened to occur during the window period when both OpenAI and Anthropic were rumored to be launching IPOs.

Background: From prophecy to admission of error

In 2025, Altman and Amodei repeatedly issued warnings about AI employment in public. Altman once told his brother that "many jobs will disappear." Amodei even more exaggeratedly predicted that AI may "eliminate 50% of white-collar jobs."

But by May 2026, both men had changed their tune.

Sam Altman: I tried it myself, AI cannot replace people

"I'm glad I was wrong," Altman admitted in a conversation with Commonwealth Bank of Australia CEO Matt Comyn. "I thought by this day white-collar entry-level jobs would be decimated, but that didn't happen."

He also talked about his personal experiment: He once tried to hand over all Slack replies and email processing to AI agents, but soon switched back to manual work - "We really care about the interaction between people. This incident made me realize that the employment picture is likely to be completely different from what we thought."

Dario Amodei: 90% automation results in 10x output

Amodei proposed a new view of the "Jevons Paradox" in an interview with Fortune a month ago: "If you automate 90% of the work, then everyone will do the remaining 10%. And this 10% will expand to 100%, and ultimately everyone's output will increase by 10 times." This is in sharp contrast to his previous prediction that "AI will eliminate half of white-collar jobs."

Sam Altman and Dario Amodei

David Solomon: I was right from the beginning

In contrast, Goldman Sachs CEO David Solomon doesn't need to change his tune. He never believed that AI would lead to mass unemployment — a position he held to in late 2025 and more recently in the New York Times Opinion Page: "The United States has a long record of creating new jobs in response to disruption...I see no reason to think that this dynamic will stop."

Key Impact

Dimensions	Changes	Impact on AI practitioners	Recommended actions
The AI jobs narrative	Shifting from 'doomsayers' to 'productivity multipliers'	Policymakers and public hostility to AI may be waning	Emphasize AI in content as a collaborative tool rather than a replacement
Enterprise willingness to adopt	CEO-level endorsement: AI is an amplifier, not a terminator	The decision-making barriers for enterprises to purchase AI tools are reduced	Emphasis on "amplifying people" rather than "replacing people" when facing enterprise customers
IPO climate	Altman/Amodei reversal coincides with IPO frontier	Moderate narratives are friendlier to investors than doomsday theories	Focus on the impact of OpenAI/Anthropic’s IPO timeline on the AI ecosystem
Direction of content creation	Declining demand for 'AI grabs jobs' content	Readers have become fatigued and are turning to practical methodologies	Reduce anxiety-based content and increase practical tutorials

Community response and industry interpretation

This story has 136 points and 115 comments on HN, and the developer community discussion is more interesting than the CEO's interview text.

It's a widely held belief that Altman and Amodei's shift in stance is highly correlated with IPO timing. The commentary pointed out that to convince the SEC and large institutional investors that the company does not have systemic risks, "AI will not destroy jobs" is a better story.

Another core discussion is: The impact of AI on employment may not be that it has not happened, but that it has not been transmitted yet. Multiple HN users cited historical analogies - both electricity and the Internet experienced curves of "overestimated short-term effects and underestimated long-term effects."

Some developers also pointed out a key blind spot: the CEO's own experience ("I tried using AI to reply to Slack, but it didn't work") may not represent enterprise-level application scenarios at all - the efficiency improvement of AI in handling customer service orders, legal review, and data entry is real, but workers in these positions do not have Altman's public voice.

Adaptation suggestions

For individual developers/freelancers:

AModei's 10x productivity theory may be closer to reality - learn to use AI tools to improve personal output instead of worrying about being eliminated
In the short term, what Altman said "human-to-human interaction cannot be outsourced" still holds true - the content generated by AI tools needs to be manually checked, which is a new value-added point

For Content Operations Team:

This report itself is good SEO material - "AI employment", "AI replacing jobs" and "Altman admitting his mistake" are all long-tail search terms
The topic will heat up again as the IPO progresses: prepare supporting Guide and Case content in advance

For business decision makers:

Solomon's argument that "the United States has historically created new jobs" is supported by solid data - companies can confidently advance their AI automation strategies
But be aware that 10x output does not mean 10x salary - changes in the division of labor structure require a redesign of the incentive system

Tool entry

If you are using the OpenAI or Claude platform to build automated workflows, you can consider incorporating the CEOs' "AI multiplier" ideas into your operational strategy—not to use AI to replace labor, but to use AI to expand the team's output ceiling. When used with n8n or ChatGPT to automate content, the review process of human-machine collaboration is maintained.

Reference sources

Internal link guidance

Want to learn how? Watch: 如何用 n8n + ChatGPT 搭建 AI 内容自动化分发系统
Real case: See how independent developers use AI tools to achieve Claude Code 48小时创业：3个月月入$9,000