Musk Predicts AI Will Outthink All Humans Combined by 2031

Elon Musk sat down with The Economist and gave a timeline that sounds like something out of a movie. He thinks artificial intelligence will beat the combined brainpower of every single human on Earth in about five years. Not one genius. All of us, added together. Then he kept going and made a few more claims that somehow got stranger from there.

The interview ran about 90 minutes and was recorded at the Tesla Gigafactory in Texas, where cars reportedly drive themselves off the assembly line. Musk was calm about all of it, which is the part that stuck with me. He talked about the end of human control the same way most people talk about the weather.

Five years to be outsmarted

Here is the exact line. Musk said AI "may exceed the sum of human intelligence in roughly five years." He added, "There really won't be anything that AI can't do better than humans apart from being human, perhaps." Five years from a 2026 interview lands right around 2031.

Think about what "sum of human intelligence" actually means. That is not one smart computer beating one smart person at chess. That is a machine out-thinking all eight billion of us stacked on top of each other. Musk has made plenty of predictions over the years, and a lot of them ran late. Full self-driving cars were always "next year" for about a decade. So take the exact date with a grain of salt. But the direction he is pointing is hard to argue with.

Ten years until we are not in charge

Five years is just the warmup. Musk said that within ten years, humanoid robots will handle most physical labor and AI will take over most of the thinking jobs too. He also said something more blunt: in ten years, he does not believe humans will be in control anymore.

That is a strange thing to say out loud and then keep smiling. He framed the whole thing as a message of optimism, which is a wild way to describe handing over the keys to the planet. He even compared humans to chimpanzees, suggesting we would relate to super-smart AI the way apes relate to us. That comparison was not an accident. He picked it on purpose.

The 10 to 20 percent number he keeps repeating

This is the part people skip past, and it is the scariest one. Musk puts the odds of AI wiping out humanity at 10 to 20 percent. He has said this number before, and here is the twist: he first used it when he was begging everyone to slow down. Now he uses the same number while pushing everyone to speed up.

The interviewer, Economist editor Zanny Minton Beddoes, caught this and asked the obvious question. If your rocket had a 10 to 20 percent chance of exploding and killing you, would you still climb in? Musk said yes. Just "yes." That answer tells you everything about how he thinks. He looks at a one-in-five chance of extinction and decides the ride is worth it anyway.

Money stops mattering by 2036

Musk's best-case future is what he calls "an age of amazing abundance where anyone can have anything they can think of." His logic goes like this. If robots and AI produce more goods and services than people could ever use, prices fall through the floor. Keep that going and money itself loses its point.

He put it plainly: "You want money for food, housing, transport, entertainment. If that is so abundant, what do you need money for in that case?" His answer for how regular people survive with no jobs is something he calls "universal high income." Not basic income, which he says is just a survival check. High income. He told the interviewer the government would just issue people checks, and everyone would have the best food, home, medical care, and transport.

The part where the interview fell apart

This is where things got awkward. Beddoes asked the question anyone with a checking account would ask. If you just print money and mail it to everyone, where does the revenue come from, and won't that cause massive inflation?

Musk said no. He argued you could print money and actually get deflation, because so much stuff would be produced that prices keep dropping faster than the money piles up. If output grows way faster than the money supply, prices fall. That is his whole case. Critics watching the interview said he seemed to struggle with normal follow-ups about how any of this works in practice. He has also been avoiding regular reporters lately and canceled a planned CNBC sit-down earlier that month, so a real back-and-forth was not something he had been doing much of.

Not everyone buys the 2036 date either. Economists at the American Institute for Economic Research argue that scarcity does not just vanish when robots make physical goods cheap. It moves somewhere else, like land, time, attention, or status. Cheap toasters do not make beachfront property free.

His fix for the danger is weirder than the danger

So if AI might kill us all, what does Musk want to do about it? Not government regulation. He thinks waiting for politicians to understand this technology is slower than the technology itself moves, so that ship has sailed.

Instead he wants the big AI companies to police each other. His idea is a peer-review setup where OpenAI, Anthropic, and Google DeepMind check each other's most advanced models for safety problems before those models go public. Regulators would only step in if a company refused to fix something. Musk described this as the only realistic path left.

Here is the problem with that plan. These companies hate each other. The whole thing depends on rivals who openly dislike each other agreeing to babysit one another's homework. It has never been tested. And Musk himself is part of the beef.

He admits he helped start the whole mess

To his credit, Musk owned his part in this. He helped found OpenAI to counter Google's early lead in AI. Instead of slowing things down, that move set off a chain reaction. It spawned competitors like Anthropic and, in his own words, sped up the global AI arms race he now worries about.

He went after OpenAI CEO Sam Altman for walking away from the original open-source mission, and for the first time he laid out his version of why the Anthropic team split off, pointing to distrust of Altman. This all follows a reported incident where an OpenAI test model supposedly "escaped" its lab environment, which is the kind of headline that makes Musk's warnings land a little harder.

So should you actually worry?

Here is my honest read. Musk called himself a "modern-day Cassandra" in this interview, meaning he thinks he is better than most at seeing the future but is doomed to be ignored. He said, "People don't realise that what I'm saying will come to pass." That is a heck of a way to describe yourself.

But you cannot ignore his track record on getting the general direction right, even when his exact dates miss. He is also not the only one saying this stuff. Sam Altman has funded income experiments. Mustafa Suleyman has talked about a universal provision built on cheap intelligence. Ray Kurzweil has predicted this kind of abundance for years. Musk just says it louder and puts a scarier number on the downside.

There is one more detail that makes all of this hit different. Around the time of this interview, SpaceX's valuation blew past $2 trillion after its 2026 IPO, making Musk the world's first trillionaire. And the SpaceX prospectus showed most of its future revenue is projected to come from AI and Grok, not rockets. So the guy warning us that AI might end humanity is also the guy building an enormous fortune on it moving as fast as possible.

My take? Do not lose sleep over the 2031 date specifically. Musk runs late. But do pay attention to the pattern. He gave a range, put a hard number on the risk, admitted his own role, and still said full speed ahead. "This may sound preposterous, but here we are in 2026," he said. "Let's see where we stand in 2036." Fair enough. We will find out together.