I would be embarrassed if OpenAI were not the first major company managed by an AI CEO.
In October 2025, Ultraman once said this.
Nine months later, he changed his tune.
On July 28, in the podcast "Invest Like the Best," Altman said people "don't really want an AI CEO," because they want to know who is making decisions in a company and who to hold accountable when things go wrong.
On July 28, Okutani (right) appeared on episode 484 of the podcast "Invest Like the Best," stating that people "don't really want an AI CEO."
Moreover, in May this year, he admitted that he had overestimated the speed at which AI would eliminate entry-level white-collar jobs, suggesting that such a "job apocalypse" is unlikely to occur.
Almost simultaneously, Jensen Huang stated during a Y Combinator entrepreneurship class: “The narrative that AI is destroying jobs is completely backwards.”
Jensen Huang said at YC Startup School 2026, "The narrative that AI is destroying jobs is completely backwards."
One controls the world's most advanced models, the other holds the computing power.
Why did they all suddenly put the brakes on the "doom of unemployment" narrative?
Why did Ultraman change his statement?
Ultraman said that people want to know who is responsible for a company’s decisions and who to hold accountable when things go wrong.
So, people don't really want an AI CEO.
He also added that, even today, most people still trust and prefer to work with real human beings. In other words, even if AI can make decisions, people may not be willing to entrust their companies to it.
What prevents an AI CEO from taking office is not insufficient intelligence, but the inability to be truly held accountable.
AI can calculate the optimal solution in a second, but it cannot bear the consequences of a decision in court, in the boardroom, or before the public.
There’s a small anecdote involving Ultraman that perfectly illustrates this point.
He once tried having AI reply to his Slack messages and emails, signing off as "This is Sam's AI," but soon switched back to replying himself.
The reason is simple: he discovered that people truly care about interacting with others.
Actually, back in that 2025 conversation with the CEO of Changxiang AI, Altman had already divided the roles: humans handle external communications, while AI handles decision-making.
This time, he drew a clearer distinction between “AI makes decisions” and “humans are accountable”:
AI can make decisions, but cannot be held accountable.
Jensen Huang: "AI Destroying Jobs"
People have it completely backwards.
Jensen Huang said everyone has it backwards about "AI destroying jobs" because they've been treating "tasks" and "jobs" as the same thing.
He proposed such a decomposition method.
A job has a purpose, and achieving that purpose requires completing many tasks; AI can take over some of these tasks, but taking over tasks does not mean eliminating the entire job.
In a job, there are also communication, judgment, coordination, review, and bearing consequences—these will not disappear along with the tasks replaced by AI.
He also cited the examples of radiologists and software engineers.
AI is taking on more and more tasks like data analysis and coding, yet the number of people in these roles is still increasing.
The numbers provided by Jensen Huang are: software engineering positions increased by approximately 10% year-over-year, while radiology positions rose by about 20% over the past few years.
Why is it growing? Because there's too much of a backlog.
Patients waiting in line at the hospital, software waiting to be developed, backlog of unprocessed requests—there’s simply too much to handle.
Once tasks are accelerated, businesses and hospitals can take on more orders and expand their operations, requiring them to hire more people to handle the remaining tasks that AI cannot perform.
Ultraman also mentioned a supporting point: a year ago, everyone said software engineers were done for, but it didn’t happen.
What has truly changed is the job itself: you no longer type out code line by line, yet it’s still unmistakably software engineering.
The task has been replaced, but the work remains. However, not all of the work is still there.
Huang also mentioned that all jobs will change, and indeed some professions will disappear entirely.
His logic holds, but it also rests on an underlying assumption: companies must use the efficiency gains from AI to expand their business and hire more people.
If market demand is that limited, or if the boss directly uses this windfall to cut jobs and reduce costs, the number of positions will still decrease.
AI hasn't eliminated entry-level positions
But the entry barrier is being removed.
The University of Maryland and LinkUp conducted a project analyzing 155 million U.S. job postings since 2018.
The results show: As of the fourth quarter of 2025, there is no evidence that AI has overwhelmed hiring demand at the overall economic level.
Top: U.S. "AI Job Intensity" (AI jobs / total jobs) rose from 0.22% in Q1 2018 to 1.13% in Q4 2025; Bottom: "Recent Graduate Job Intensity" rebounded from 11.7% in Q4 2022 to 12.6% in Q4 2025, with 532,000 job postings targeting recent graduates in Q4 2025.
The most counterintuitive is the recent graduate category.
We’ve heard “AI will eliminate entry-level jobs” so often over the past two years that it’s become a cliché—but this data shows the proportion of positions explicitly targeted at recent graduates has actually increased, rising from 11.7% in Q4 2022 to 12.6% in Q4 2025.
In the fourth quarter of 2025 alone, there were 532,000 such job postings, a 63% increase compared to the first quarter of 2018.
Even in fields most expected to be disrupted by AI—mathematics and computer science—entry-level positions have increased by 55% compared to the first quarter of 2018.
Seeing this number, the claim that "entry-level positions are disappearing" is at least unfounded in terms of total numbers.
There's another counterintuitive observation in the report:
Younger, less experienced employees are more likely to benefit from AI tools, as AI excels at instantly providing them with vast amounts of expertise, making them appear as "cost-effective and highly useful" early adopters in their employers' eyes.
The step at the entrance
Narrowing
The total supply hasn't collapsed, but that doesn't mean there's no cost: there's something these impressive numbers can't hide.
How did a young person grow up in the past?
Often start with the most standardized entry-level tasks: data entry, basic coding, fundamental analysis, and organizing documents.
Although these tasks may be dull and repetitive, it is through doing them over and over again that people gradually build up experience and judgment, honing skills that cannot be replaced.
And these standardized tasks are precisely the ones AI will take over first.
The entry point to entry-level positions is narrowing, and this is the most immediate pressure facing recent graduates and newcomers to the industry.
Your true moat
When we combine these two statements about Ultraman and Jensen Huang, we find:
The more capable AI becomes, the more the value of human work shifts upward—from “getting the job done” to “taking responsibility, building trust, and setting goals.”
Machines can gradually replace tasks one by one, but they still cannot fill the gap of "who makes the final decision and who takes responsibility for the outcome."
Instead of worrying about whether AI will take your job, ask yourself a different question: How much of your work requires your signature and responsibility? How much of it must be done by you personally for others to accept it?
The parts that AI cannot replace are your true moat.
Reference materials:
https://open.spotify.com/episode/5bFBUf3X8QXaDFrve1Be5s?utm_source=chatgpt.com
https://conversationswithtyler.com/episodes/sam-altman-2/?utm_source=chatgpt.com
https://whbl.com/2026/05/26/openais-altman-says-ai-unlikely-to-lead-to-jobs-apocalypse/?utm_source=chatgpt.com https://youtu.be/I4B37S1dyQQ
Edited by: Yuan Yu
This article is from the WeChat public account "New Intelligence Yuan," authored by ASI Revelation.