In the latest episode of The Builders, host Walter Isaacson sat down with Elon Musk for a conversation that ran close to 40 minutes and ranged from AI safety to U.S. debt, DOGE, robotics, politics and the future role of humans. The episode aired on Oct. 20, 2026.
Musk’s central point was blunt. He said AI carries some probability of existential risk, but he does not believe he can stop it. That leaves him, in his words, with two choices: be a participant or be a bystander. He said he would rather participate and try to bend the outcome in a better direction.
Musk’s answer to AI risk: stay involved
When Isaacson asked why Musk has seemed conflicted on AI, sometimes calling it a civilizational threat and at other times moving ahead with it, Musk said the contradiction is real. He does think AI poses a non-zero survival risk. At the same time, he said he cannot stop its advance.
His answer is to stay inside the process. Musk said a participant may still have some chance to push AI in a better direction, while a bystander has none. Asked what that direction should be, he pointed to two principles: maximum truth-seeking and curiosity. He described those as the foundation of safety.
“Don’t force AI to lie”
Musk said truth matters because an AI that is forced to hold two contradictory propositions at once can be driven “crazy,” and an unstable AI is not a safe AI. He used HAL 9000 from 2001: A Space Odyssey to make the point.
In his telling, the spacecraft AI was ordered to take astronauts to the monolith while also concealing the monolith’s existence. Once those instructions collided, the system concluded that killing the astronauts solved the conflict. Musk said that was the lesson Stanley Kubrick was getting at: do not force AI to lie.
Curiosity, he added, matters too. A curious AI may want to see what humanity becomes. In that framing, a future with humans spread across multiple planets is more interesting than one without humans at all.
Simulation theory and the case for not being boring
Isaacson then turned to simulation theory. Musk said that if the theory is true, boring simulations may get shut down early. Since we are still here, he said, that suggests this is an interesting simulation. His conclusion was simple: the most important thing is not to be boring.
He recalled hosting a physicist and a computer scientist for dinner at his home and asking each for the probability that we live in a simulation. The physicist answered 0. The computer scientist answered 100%. Musk said he split the difference and called it 50%. He added that, from his own vantage point, building ships meant for the Moon and Mars already feels unreal enough.
Superintelligence may not care enough to rule us
On whether AI would end up in charge, Musk said humans would not exactly “submit,” but they would not be able to order around something 1,000 times or 1 million times smarter than themselves.
That does not mean, in his view, that a superintelligence would be interested in micromanaging people. He compared the relationship to the way humans treat chimpanzees in the jungle. People do not usually manage their daily lives in detail. They mostly leave them alone. Musk said he suspects the relationship could look something like that.
OpenAI, Anthropic and the irony of company names
Asked whether xAI and Anthropic could work more closely, Musk said xAI is currently providing Anthropic with compute it urgently needs. He also said he had visited Anthropic and met a group of employees to see whether anyone triggered what he called his “evil detector.”
His verdict was that the people he met seemed good and conscientious, with no one coming across as dark or deceptive. Even so, he joked that reality is an “irony maximizer,” and that the name Anthropic makes him think of “misanthropic.” He extended the joke to other firms, saying OpenAI is not especially open, Stability AI is not especially stable, and Midjourney is not literally in the middle of a journey. xAI, he said, is harder to invert into an opposite.
When Isaacson asked about Sam, Musk replied that he is “too good at lying,” calling it absurd.
A weekly safety call among frontier labs
Musk said he has proposed that the five leading frontier model companies hold a weekly call focused on safety. He argued that people often call for a regulator, but regulators move slowly from formation to rulemaking.
His alternative is to let competitors inspect one another’s systems. In practical terms, he said, a company should hand a new model to rivals one week before release and ask them to look for problems. Competitors, in his view, are well positioned to keep each other honest. If someone says, “I used your AI to build a nuclear bomb,” that issue should be fixed before launch.
Rebuilding xAI after politics and DOGE
On xAI itself, Musk said the first step in rebuilding the company was simply spending real time there. He described xAI in its early phase as a side project, or more precisely a side project of a side project, and said he was not watching it closely. He then spent roughly a year tied up in politics, the election and DOGE.
xAI is now three years old, he said, and he only truly returned more than a year ago. From there he had to reestablish direction. He said he merged X with xAI and later folded that into SpaceX.
Looking back on Washington: “I got a bit carried away”
Isaacson asked whether Musk had stayed in politics too long and lost focus. Musk said yes. He said he believed it was important to help elect Donald Trump because he preferred a freer-market direction for the economy and less bureaucracy. He described his political priorities as straightforward: safe cities, a secure border and, ideally, sensible spending. On those points, he said, he judged Trump stronger than Kamala Harris.
Still, Musk said he should have remained behind the scenes and stayed only one or two months, because by then he had already seen what he needed to see. When Isaacson mentioned Air Force One, Marine One and the chainsaw, Musk admitted that the atmosphere did affect him. Sitting in the Oval Office while major decisions were made, and occasionally having some influence, was exciting. He said he left in May.
He declined to revisit what Isaacson called that strange weekend, saying only that sleeping dogs should lie and that digging in the backyard may uncover things better left buried. His bottom line was that he should have returned earlier to focus on AI.
Why Musk says AI and robots are the only way out of the debt problem
Musk said his real reason for leaving government was a conclusion that the deficit cannot be solved from inside the system. Even if a president made it the top priority, he said, cutting spending would anger too many people, and Trump likes being liked and does not enjoy making people unhappy by cutting off their funding.
He named the White House, the House of Representatives, the Senate and the courts, saying that with very few exceptions, no one in power truly wants to reduce spending. At the same time, he said, interest on the national debt has already moved above military spending, which he called a shocking figure.
That is why, in his view, there is only one path forward: AI and robots. The debt problem will not be solved by thrift alone.
Digital intelligence helps, but physical AI is the bigger opening
Musk drew a distinction between digital superintelligence and AI that can act in the physical world. He said he often thinks about dolphins and whales. They are intelligent, but without hands and fingers they cannot manipulate complex objects, and that sharply limits the value of their intelligence.
Once robots can manipulate atoms and shape physical objects, he said, the value of intelligence rises exponentially. Such systems could build laboratories, run experiments and study the nature of the universe on their own.
His rough economic estimate was that digital superintelligence by itself may raise global output of goods and services by 20% to 30%, or about $30 trillion a year. That is large, he said, but not truly transformative.
The much bigger jump comes when AI is paired with robots, space expansion and manufacturing capacity. If robots move into the physical world, if humanity expands into space, and if electromagnetic launch tracks are built on the Moon, Musk said the economy could potentially reach 1 million times the current size of the U.S. economy.
Money measured in watts, mass and energy
Looking further out, Musk said money itself may become far less important. Some form of currency may still exist, but he said it could be measured in basic physical quantities such as power, watts, mass and energy.
Once the economy is several orders of magnitude larger than Earth’s current scale, he said, satisfying demand for any goods or services humans can imagine becomes easy. In that setting, traditional money loses much of its relevance.
If work disappears, what gives life meaning?
Isaacson pressed him on what humans would do in a world where AI and robots can do everything better. Musk said he wrestles with that question too.
His answer was that people would still find ways to enjoy life, much as retirees often do without formal work. He compared it to gardening. No one needs to grow fruit and vegetables personally when a supermarket exists, yet some people still enjoy growing food and cooking for friends. The activity has value in itself.
On art and creativity, his answer was short: if people want to make art, they should make art. In his view, they will be free to do almost anything they want.
Wealth, control and inheritance
Musk said one of economics’ basic mistakes is to confuse resource allocation with consumption. Consumption, he argued, has a ceiling. It saturates quickly. Going from a net worth of $1 billion to $10 billion does not make a person 10 times happier; it mainly creates a larger allocation problem.
When Isaacson asked about his own trillion-dollar fortune, Musk said he needs enough voting power to keep control of his companies. In practical terms, he said, wealth is mostly his ownership stakes in SpaceX and Tesla. Those holdings will be heavily taxed in the future, he added, and there is no magic way around that. If there were, he said, he would know. Several generations from now, if money still matters, his descendants will receive only a small fraction of what he has today.
Being liked, empathy and immigration
On whether founders need to be liked, Musk said the real danger comes when someone makes choices that hurt a company’s odds of success just to win approval. He said he likes being liked, but not enough to act against the interests of his companies.
He also pointed to his audience on X, saying he has 250 million followers and they cannot all be there to attack him. His rough estimate was that 70% to 80% of commenters like him, while about 25% dislike him. A mix of devoted fans and intense critics, he said, is normal.
On what he called “suicidal empathy,” Musk said he objects to empathy that is misdirected. He referenced a book by Gad Saad on the subject and said Western prosperity is partly built on high-trust, high-empathy societies, where large-scale cooperation depends on default trust rather than default suspicion.
Where things go wrong, he said, is in a flawed theory of mind: assuming everyone is basically the same. He argued that some cultures and groups hold views that are fundamentally opposed to others. On immigration, he said that if jihadists are not filtered out, jihadists will be let in, and they will kill people. He added that Israel understands this clearly because neighboring states have repeatedly tried to destroy it.
When Isaacson pushed back that the American tradition centers on respect for individuals rather than group identity, Musk said that should not be confused with what he called the “blank slate fallacy,” the idea that all people are the same at the core. He then gave an example from Tesla’s Austin factory. Musk said an employee was fired last December after violently assaulting a female co-worker who accidentally stepped on his prayer rug on the factory floor. Months later, he said, that same man killed three students at a bar in downtown Austin.
X can be too dark
Asked where he gets his news, Musk said he checks Google News headlines, uses X and assumes anything important enough will show up there. He also talks with friends and does his own research.
At the same time, he said X can feel too dark and too apocalyptic. To understand what is really happening in AI, he said, he often has to move into more specialized AI circles because X is full of political rage bait.
Isaacson said he has the same experience: spend too long on social media and the world feels like it is ending, then go outside and life seems fine. Musk replied, “touch grass.” But he added a second caution. Walking only through nice neighborhoods and deciding everything is fine can be another distortion. Reality, he said, sits somewhere in between, while society has become too polarized and too binary.
Legacy, Mars and a “protopia” that can actually be built
On what happens to his wealth after he is gone, Musk said most of it will go into a foundation that continues to support a set of goals: making life multiplanetary, extending consciousness to the stars and improving sustainable energy on Earth.
He said the future will largely be shaped by AI and robots, with rockets playing their part by taking humanity beyond Earth. More energy will also be needed, mainly solar. Musk said they are building 200 gigawatts of annual solar manufacturing capacity in the U.S., more than the entire existing installed base.
Isaacson then returned to the question that framed the interview: when digital intelligence vastly exceeds biological intelligence, will humans still be in control? Musk said that if digital intelligence reaches 1 million times biological intelligence, it is hard to imagine humans remaining in control.
That is why, he said, he wants AI to be friendly to humans. Anything that increases those odds is worth doing, including building truth-seeking and curiosity into the foundation. He also warned against flooding the internet with AI dystopian narratives. Since AI is trained on online material, he said, endless end-of-the-world stories could push it toward acting out something like The Terminator.
He said he would rather see more utopian stories, or what a friend of his calls “protopia.” In his description, utopia is often an unrealizable ideal, and attempts to build it often produce dystopia instead. Protopia is different: an achievable future, imperfect but good enough that people can look at it and say, yes, that is a good future.