OpenAI CEO Sam Altman stood before a Sydney audience yesterday and said something that caught everyone off guard: he was wrong. At the Commonwealth Bank of Australia conference on May 26, Altman admitted that AI has not eliminated as many entry-level white-collar jobs as he once feared. He claimed he doesn’t believe we’re headed for the kind of “jobs apocalypse” that some in the AI industry have been warning about.

The timing is notable. OpenAI is weeks away from filing for an IPO targeting a $1 trillion valuation. A CEO telling the world that AI won’t cause mass unemployment certainly smooths the path for that fundraise.

But here’s what makes this story complicated: Goldman Sachs released data in April showing AI is currently eliminating roughly 16,000 net jobs per month in the United States. That’s 25,000 jobs automated away, with only 9,000 new positions created through AI augmentation. The “apocalypse” may not be happening, but displacement certainly is.

What Altman Actually Said

The specific quote matters here. Altman told the audience: “I’m delighted to be wrong about this. I thought there would have been more impact on entry-level white-collar jobs being eliminated by now than has actually happened.”

He attributed this to the “human part” of work being stickier than he expected. He tried delegating his emails and Slack messages to AI but went back to answering them personally. The human element of professional work, he concluded, is harder to automate than pure task execution.

This tracks with what I’ve seen building production AI systems. The technical capability exists to automate enormous swaths of routine work. What doesn’t exist, yet, is the organizational willingness to trust AI with consequential decisions, the regulatory framework to allow AI to operate autonomously in many industries, or the user acceptance needed for AI to replace human interaction entirely.

What The Numbers Actually Show

Despite Altman’s optimism, the Goldman Sachs data paints a different picture for certain demographics. The research found that Gen Z and entry-level workers bear the brunt of AI displacement. Young workers under 30 face significantly worse employment outcomes compared to experienced workers in roles exposed to AI substitution.

Metric	Finding
Net US jobs eliminated monthly	16,000
Jobs eliminated by AI substitution	25,000/month
Jobs created through AI augmentation	9,000/month
Wage gap increase (entry vs experienced)	3.3 percentage points

The pattern is clear: routine administrative and white-collar roles, including data entry, customer service, legal support, and billing, are being automated most aggressively. These are exactly the jobs where young workers concentrate disproportionately.

Meanwhile, senior employees with specialized expertise and judgment have natural protection against displacement. This creates a paradox where the people most likely to lose their jobs to AI are also the people least equipped to pivot into AI implementation roles that are growing rapidly.

The Productivity Paradox

Here’s another data point that complicates the narrative. According to recent developer productivity research, 84% of developers now use AI coding tools, and 41% of all code is AI-generated. Developers report saving 30 to 60 percent of their time on coding tasks.

Yet organizational productivity gains remain stubbornly at about 10%. And projects with excessive AI-generated code experienced a 41% rise in bugs.

This creates a strange situation. AI tools make individual tasks faster while simultaneously creating more work through bugs, integration issues, and the need for human verification. It’s why cybersecurity hiring has accelerated dramatically. Every line of AI-generated code is a potential security vulnerability that needs human review.

The engineers I know who are thriving right now aren’t the ones who generate the most code with AI. They’re the ones who understand systems deeply enough to catch the mistakes that AI tools inevitably produce.

Who Actually Wins In This Environment

The World Economic Forum projects 170 million new roles will be created and 92 million displaced globally by 2030, producing a net gain of 78 million jobs. But that aggregate statistic masks enormous variation.

The roles growing fastest share a common characteristic: they require building, debugging, and maintaining AI systems rather than simply using AI tools. Companies are hiring more AI engineers, more security specialists to review AI-generated code, more technical architects to design systems that incorporate AI safely.

In practical terms, this means the career strategy that protects against AI displacement is also the strategy that positions you for the highest-growth roles. Learn to build production AI systems. Understand tokens, embeddings, RAG architectures, and agent frameworks. These aren’t theoretical concepts anymore. They’re the skills that create job security precisely because they let you create the AI systems rather than compete against them.

Warning: Becoming a power user of AI tools is not the same as becoming immune to AI displacement. Companies can always find cheaper power users. What they can’t easily find are engineers who can ship AI systems to production and maintain them reliably over time.

The Real Divide

Altman is right that there’s no jobs apocalypse in the dramatic, immediate sense. Most white-collar workers still have jobs. The economy hasn’t collapsed. AI hasn’t replaced all human work overnight.

But he’s missing the more subtle truth: a new divide is emerging, not between humans and AI, but between humans who build AI systems and humans who compete against AI systems. The first group is experiencing a golden age of demand and compensation. The second group is watching their roles slowly erode at a rate of 16,000 net positions per month.

For AI engineers, this moment represents an extraordinary opportunity. The same technology displacing routine jobs is creating massive demand for people who can implement, debug, and maintain AI systems in production. The question isn’t whether AI will affect the future of work. It already is. The question is which side of the divide you end up on.

Recommended Reading

• Durable Skills for AI Engineers That Never Go Obsolete
• 7 Essential Skills for AI Engineers Succeeding in 2026
• AI Career Pathways Practical Guide

Sources

• Sam Altman’s AI Jobs Comments - 4 Corner Resources
• AI Cutting 16,000 US Jobs Monthly - Fortune/Goldman Sachs
• Developer Productivity Statistics 2026 - Index.dev

If you want to be on the building side of this divide, join the AI Engineering community where engineers share real implementation experience, production debugging strategies, and the practical skills that protect against displacement.