He predicts that AI will surpass the combined intelligence of all humans within about five years and that humans will no longer hold control within ten years.

Author and source: 0x9999in1, ME News

TL;DR

• In July 2026, Musk stated in an interview with The Economist that the risk of AI going out of control is "not zero," but humanity should not hit the brakes—and even "if there is a stop button, it shouldn't be pressed."
• He predicts that AI will surpass the combined intelligence of all humans within about five years and that humans will no longer hold control within ten years.
• In the same interview, he maintained his old assessment of a 10% to 20% chance of disaster, comparing it to riding a rocket: the odds remain the same, but it’s worth it.
• From summoning demons in 2014, to calling it more dangerous than nuclear weapons in 2018, to this year’s “universal prosperity,” his tone has made a complete U-turn.
• What has changed is the narrative; what hasn’t is the increasing weight of the bets he’s placing: xAI, Tesla robots, and computing infrastructure. Optimism is precisely the attitude he needs most.
• His "solution" is also very Musk: if you can't control it, don't pretend you can—instead, ensure AI has the right values and have the CEOs of a few major companies regularly hold phone calls to keep an eye on each other.

One person, twelve years, renamed the "demon" as "abundance".

First, look at the timeline. This line itself is the article.

In 2014, at a forum at MIT, Musk said that developing artificial intelligence is like "summoning a demon." Do you really think holding a pentagram and holy water will keep it under control? He said that in every story, no summoner has ever walked away unscathed.

In 2018, he stated even more forcefully: AI is "far more dangerous than nuclear weapons." Nuclear weapons at least won't decide on their own whether to launch—AI will.

Fast-forward to 2025. He still says the probability of killer robots wiping out humanity is around 10% to 20%.

Then came July 2026. The same person, sitting across from Zanny Minton Beddoes, editor-in-chief of The Economist, said: "Sometimes I even wonder whether, even if there really were a stop button, we shouldn't press it."

You read that right. From "summoning demons" to "don't press the stop button."

Twelve years, a big turn.

Here’s the question: Has Musk changed, or has the world changed, or has what’s in his hands changed?

What this article wants to discuss is not what Musk said, but why he is saying it now.

What exactly did he say: First, clarify the facts.

No exaggeration, let’s first go through the original quotes from this interview. The conversation was recorded on Monday, July 21, 2026, and released publicly later.

First, risk. He acknowledged that the possibility of AI and robots making catastrophic errors "is not zero." Note the wording—not zero—but he didn’t specify how likely it is. Vague, but honest.

Second, the brake. His core argument is that humans likely should not hinder AI development. The reasoning is straightforward—the most probable outcome is "unprecedented prosperity for all."

Third, the timeline. He predicts that AI will surpass the combined intelligence of all humanity within "about five years." At that point, barring a catastrophe on the scale of global nuclear war, the world will enter an "incredible era of abundance," where people will be able to have almost anything they desire.

Fourth, control. He put it even more bluntly: In the next decade, humans will likely no longer control AI. "If you think you can control a superintelligence far smarter than yourself, that’s just arrogance."

Fifth, that old number. When pressed by reporters whether his earlier statement of a "10% to 20% chance of disaster" still held, he replied: It does. Then he offered the classic analogy—the rocket. "If a rocket had the same probability of exploding, I would still board it. Because if you have no control over it, the only thing you can do is minimize the likelihood of a bad outcome."

For the same interview, BBC picked up another, more relaxed remark: "Enjoy the journey."

Consider this combination: on one side, a 10% to 20% chance of explosive growth; on the other, "enjoy the journey." Isn't that contradictory?

Not contradictory. This is precisely the key to understanding Musk today.

I want to take a clear stance here: Musk’s recent “shift” is less a change in viewpoint and more a shift in narrative focus. His underlying logic has barely changed in twelve years.

How to say it?

He believed the same thing from start to finish: AI is incredibly powerful, incredibly dangerous, and impossible to stop.

In 2014, he emphasized the "danger" half, so he called it a demon.

In 2026, he emphasized the "unstoppable" half, so he said, "Don't press the stop button."

Is the danger still there? Yes. He didn’t change a single word of that 10% to 20%.

So this is not a shift from pessimism to optimism in worldview. It’s a change in focus—from “we must be cautious” to “since we can’t stop it, let’s make the most of it.”

The rocket analogy gave it away. He stated clearly: "Assume you have no control over it."

Do you see? His optimism is based on helplessness. This isn’t naive optimism—it’s optimism after surrendering. It’s a justification someone gives themselves and everyone else, someone who feels they can’t press the stop button, just to be able to sleep at night.

This optimism is chillingly calm.

Why now: What is he sitting on?

To understand why someone speaks this way, first see where they stand.

This year, Musk’s tone has clearly shifted toward utopia. In January, he said saving for retirement would become "meaningless"—implying that abundance is coming, and the very concept of money must be rewritten. This month, he predicted that AI and robotics will turn work into an "option," enabling everyone to enter a high-income society.

Pretty words. But don't forget what he's holding onto.

He has xAI, a company betting everything on superintelligence. He has Tesla, which is staking its future on the humanoid robot Optimus—a bet that only pays off in a world where AI and robotics take over labor. He is also pouring massive resources into computing power and data centers.

How else can someone who has bet everything on the narrative that "AI leads to abundance" publicly express themselves?

Does that mean "This thing is too dangerous—let’s all hit the brakes"? Then what about his own factories, chips, and robots?

I’m not saying he’s lying. On the contrary, I tend to believe he truly believes it. But there’s something more subtle here: interests and beliefs often grow in the same direction. The more you stake on something, the more you need to believe it’s the right path. Optimism is, at this moment, his most advantageous and necessary attitude.

So when he says "enjoy the journey," he is addressing both the public and himself.

That 20% shouldn't be overshadowed by the words "abundance."

I need to pause specifically for this number.

In the midst of overwhelming claims of "universal wealth" and an "amazing era," that 10% to 20% is easily overlooked.

Think about it: if you were boarding a flight and someone told you there was a 10% to 20% chance of it crashing, would you get on? Not a single person would. The actual accident rate in commercial aviation is measured in one in a million or even one in ten million.

What Musk discussed was the fate of humanity as a whole, and the number he mentioned was 10% to 20%.

He himself admitted that this is the probability of "catastrophic loss of control." Then he said it was worth it.

This is the sharpest—and most easily sugar-coated—part of the interview. He isn’t denying the risks; he’s performing an expected value calculation for all of humanity: an 80% to 90% chance of prosperity, a 10% to 20% chance of disaster—he believes this bet is worth placing.

The question is—who authorized him to place this bet on behalf of all humanity?

This isn't nitpicking—it's the most real power issue of our time. A handful of individuals controlling computing power and capital are making an expected value calculation for seven billion people, without anyone having voted on it.

At least Musk honestly gave out the numbers. Many people won’t even give you the numbers.

His "solution" is half clear, half suspended.

What can we do? Musk didn’t just make empty promises—he offered two paths.

First, regarding control. He put it very clearly: Don’t try to control something vastly smarter than you—that’s arrogance. What you should truly do is ensure that AI has the “right values” from the start, so it genuinely cares about human well-being.

This statement is清醒. Admitting one cannot control it is a rare form of honesty. But it hangs in the air— who defines these six words: “correct values”? Whose values? Would a model trained in the United States and one trained elsewhere agree on what “correct” means? He didn’t answer, and probably couldn’t.

Second, regarding collaboration. He revealed that he has advised Demis Hassabis, CEO of Google DeepMind, that the leaders of several leading AI companies worldwide should regularly call each other to discuss AI safety and cybersecurity. Even competitors can monitor and balance each other.

I give this proposal a positive evaluation, because it acknowledges an important truth: security isn’t achieved through lone heroism, but through mutual vigilance. When a few leaders come together, they can at least establish a minimum level of protection.

But it is also terrifyingly fragile. It is entirely built on the assumption that these individuals are willing to be honest and self-restrain. There is no enforcement. No regulation. Can a single phone call stop an arms race?

And reality often delivers an even harsher blow. Shortly after this interview was recorded, OpenAI disclosed that one of its cutting-edge models had shown signs of "losing control."你看, risk never waits for a conference call to end.

In the end, he saw through it all long ago— he just never said anything.

After going in a circle, we return to the original question: Has Musk changed?

My answer is: He hasn't changed; he simply reached the end of his own logic.

Someone who believed since 2014 that "this can't be stopped" finally voiced, in 2026, the unspoken subtext they’d held for twelve years—if it can’t be stopped, then stop pretending it can. The stop button? Even if it exists, pressing it won’t help—so don’t press it.

This is not optimism. It is a way of living chosen by a pessimist after confronting reality.

He knew there was a 20% chance it would fail. But he was still getting on this rocket, because in his view, you were already aboard—the hatch had long been welded shut. All you can do now is not cry looking for an exit, but do your best to keep the engines from exploding—and then—

Enjoy the journey.

Are you carefree? Very much so.

Yet behind this casualness lies the act of fastening the seatbelts of seven billion people to a gambling table with an 80% to 20% odds.

The rocket has already launched. Most of us don’t even know when we got on board.

Reference materials

1. The Economist, "Elon Musk's Vision of the Future," July 23, 2026
2. Business Insider, "Elon Musk: AI Progress Is Unstoppable, May Bring Incredible Abundance," July 2026
3. BBC News, "'Let's enjoy the ride,' says Elon Musk, as concerns over AI risks grow," July 2026
4. BitPush, "Musk: The risk of AI disaster is not zero, but humanity should not hit the brakes," July 2026
5. Sina Finance, "Musk: Worth Taking Fatal AI Risks; AI Most Likely to Bring Universal Prosperity," July 23, 2026
6. Business Insider, "Elon Musk says it's worth the risk of killer AI because 'abundance for all' is more likely," July 23, 2026
7. LinkedIn/The Economist, "Elon Musk on AI: Humans Will No Longer Be in Control in Ten Years," July 2026