Musk's latest interview: AI could potentially wipe out humanity, but like a rocket with a 20% chance of exploding, I would still get on and enjoy the ride!
```
In the face of the unstoppable AI wave, Elon Musk has presented an aggressive timeline and pragmatic regulatory plan: he predicts that AI will surpass human intelligence within five years, and top companies in the industry should immediately establish a mechanism of “mutual audit” of models to manage safety risks.
Recently, Tesla CEO Elon Musk was interviewed by Zanny Minton Beddoes, editor-in-chief of The Economist. In this interview, Musk not only gave his latest assessment of the timeline and potential risks for artificial intelligence (AI) surpassing humans, but also, for the first time, detailed his regulatory vision for the top AI large model development companies today, and did not shy away from talking about the grudges among key figures in Silicon Valley's AI giants.
AI Will Surpass Human Intelligence Within 5 Years, Probability of Human Extinction is 10% to 20%, He Has No Choice But to "Get On Board"
The market is highly focused on the timetable for AGI. To this, Musk gave an extremely clear and aggressive prediction: “About five years from now, artificial intelligence may surpass human intelligence.”
With the explosion of intelligence comes the loss of control. When asked whether humans will still be able to control the situation in ten years, Musk bluntly said this is “unlikely.” He used a harsh but vivid biological analogy to explain his logic: “If the intelligence gap between AI and humans is much larger than the gap between AI and chimpanzees, then it’s hard to imagine chimpanzees would be dominant. We are actually just evolved versions of chimpanzees; not long ago, we were eating bananas in the jungle.”
Musk has repeatedly warned that AI could bring about the end of humanity. In this interview, when asked whether the probability of killer robots causing human extinction remains at 10% to 20%, Musk did not deny the risk, but chose an optimistic attitude tinged with fatalism.
He made a strongly visual analogy with rocket launches. When the host asked, “If your rocket had a 10% to 20% chance of exploding and killing you, would you still get on it?” Musk answered affirmatively: “Yes. All we can do is try to minimize the probability of bad outcomes.”
He further explained the reality behind having to “get on board”: “The development of AI and robotics seems unstoppable — even if I wanted to stop it, I couldn’t.” He believes that since it can’t be stopped, the best choice is to stay involved to influence its direction: “The most probable result is enormous benefit for everyone.” Excluding extreme events such as global nuclear war, Musk predicts that by around 2036, humanity will likely usher in an “era of great material abundance, where everyone can have everything they want.”
Calls for Establishing AI “Mutual Audit Mechanism”: Use Business Competition to Keep Things Honest
With AI giants like OpenAI, Google, and Anthropic engaging in an arms race over foundational models, how to regulate effectively has become a market focus. Musk suggested a very down-to-earth solution in this interview: let competitors audit each other's models.
Musk revealed that he had a long conversation with DeepMind founder Demis Hassabis, and believes the most urgent action is to “have the leading AI companies at least meet or communicate regularly, discussing any safety-related issues — for example, every week or two.”
Musk noted that government officials lack strong technical backgrounds, making it difficult to judge whether a cutting-edge model should be released. Therefore, business rivalry among giants should be leveraged to ensure safety: “For competitors, as long as they get a week or two to review a new model, they can spot its issues, which incentivizes all competitors to remain honest.”
The core logic of this mechanism is that “fiercely competing companies can realize that something might be a safety risk, so they won’t hesitate to point it out, and recommend delaying the release of a rival’s model.” Only if a company refuses to deal with risks should the government step in. Musk stressed this system must be established immediately; the so-called ‘six months to build it’ proposed by the industry is too long, since there are now multiple AI breakthroughs every single day.
Harsh Criticism of OpenAI’s Betrayal of Its Original Intent, Yet Hopes to Set Aside Personal Disputes for the World’s Benefit
When talking about barriers to cooperation in the AI industry, Musk made no secret of his dissatisfaction with OpenAI’s current CEO Sam Altman. This frank statement also reveals the root of the lack of trust between top AI labs to the public.
“I don’t like Sam Altman because OpenAI was originally meant to be an open-source AI company for the world. But somehow, it became an $800 billion closed-source, for-profit company.” Musk bluntly said this completely violated his original intention when he donated.
At the same time, Musk also revealed inside stories about Anthropic (currently one of the most valuable AI startups, founded by former OpenAI core members): “Dario [Anthropic CEO] doesn’t like Sam Altman either. In fact, the reason the Anthropic team left OpenAI was that they didn’t trust Sam Altman, otherwise they would still be working at OpenAI now.”
Despite the intense personal resentment among the giants, Musk showed a gesture of compromise at the end of the interview. He said the Anthropic team are “principled people who care about the world’s future.” To push forward the establishment of AI safety auditing mechanisms, Musk appealed: “For the good of the world, let’s put personal differences aside.”
Transcript
Musk
I think that in about five years, artificial intelligence may surpass human intelligence. Five years — that is about the predicted timeline. So, except for being human itself, AI can indeed do things better than people.
Host
So, what will life be like in the future?
Musk
The most likely scenario is that we’ll reach an era of great material abundance, where everyone can have whatever they want. This may sound a bit far-fetched, but, well, we’re already living in 2026. Let’s see where things stand in 2036.
Of course, there may be some factors that obstruct this goal, like a massive global nuclear war or such events. However, assuming World War III doesn’t happen, I think we’re heading toward a period of material abundance. So, perhaps that can count as a signal of optimism and excitement about the future.
Host
Are you sure that in ten years, mankind will no longer be able to control the situation?
Musk
I don't think that's very likely. If the intelligence gap between AI and humans is much bigger than between AI and chimpanzees, then it’s hard to imagine chimpanzees being in charge. We are essentially just evolved chimpanzees. Not long ago, we were eating bananas in the jungle, and even now, we still eat bananas.
Host
Do you think that the probability of killer robots causing the end of humanity is still 10% to 20%? You seem pretty relaxed about it now. Have you changed your mind?
Musk
I still think AI and robots pose a risk — that risk is not zero. But I have to take an optimistic attitude. Sometimes, regarding the development of AI and robots, I even think that even if there were a stop button, we might not necessarily want to press it — because the most likely result is enormous benefit for everyone. I realize that now everyone seems to believe this development is inevitable. Let’s hope for the best and move forward. The development of AI and robotics seems unstoppable; even if I wanted to stop it, I couldn't.
Host
You think there’s still a 10% to 20% chance of killer robots causing human extinction?
Musk
That’s really a bad situation. In the long run, all of us are headed for death anyway. According to the standard model of physics, the end of the universe will be a long and boring process — like a thin, cooling gray soup — the so-called ‘heat death’ of the universe. So, if heat death is indeed the universe’s final outcome, then all of this is just part of the journey.
Host
If you thought your rocket had a 10% to 20% chance of exploding and killing you, would you still get on it?
Musk
Yes. All we can do is try to minimize the probability of a bad outcome. I was always reluctant to get involved in AI or related fields. At first, OpenAI existed to counterbalance Google, because Google nearly monopolized AI development. Now, Anthropic has become a leader in AI. These efforts actually accelerated AI development, which was never my intention — it all seems to be pointing toward AI’s further development. So I thought, okay, you can be sad about it, or join in — and obviously, you’ve joined.
Host
If you can’t beat them, join them. As I understand it, some other leaders in the field of model development are also looking for ways to reduce risk. Just last week, Demis Hassabis made a proposal about establishing a public-private regulatory body. Do you think you should be more actively involved? Are there measures we can take to control these risks?
Musk
Yes, my advice to Demis is — we talked for hours before he published his article. We think the most urgent thing is for leading AI companies to at least have regular meetings or exchanges, to discuss any safety-related issues, every few weeks. Why haven’t you done this already? I mean, you don’t all like each other — after all, you’re competitors, and everyone wants to be number one.
Musk
But I think this kind of competition can keep people honest. For government officials who don’t have a strong technical background or don’t lead the frontier of AI, judging whether a model should be released is truly difficult. But for competitors, if you give them a week or two to review a new model, they’ll spot issues, which incentivizes all competitors to stay honest. In this way, creators of models can check each other’s work. If something worries them, they can tell the government, which can then judge if there’s an issue that needs solving — and if the company doesn’t address the risk, that’s when the government should step in and take action.
Action should be taken immediately. People keep telling me that such a system must be established within six months, since model improvement is accelerating. But six months is really too long — it should be faster. We’ve seen so many impressive AI breakthroughs — sometimes several in a single day, and there were many just last week.
Host
So you need to get together now and get this done — sounds like that’s what you’re telling me.
Musk
Yes, I suggest at least holding some informal meetings regularly, maybe weekly or biweekly. That way there’s a week or two of advance access to the relevant content.
Musk
That’s why I think this incentive mechanism works — highly competitive companies can spot something that might be a safety risk, and will not hesitate to point it out, and recommend delaying the release of a rival’s model.
Host
You don’t really like each other, and there’s a lack of trust. You often attack each other on social media. So do you think you can do this with people you don’t like? There really is personal animosity among these leaders.
Musk
I don't like Sam Altman, because OpenAI was originally supposed to be an open-source AI company, belonged to the whole world. But somehow, it became an $800 billion closed-source, for-profit company. I think this completely violated my original purpose in donating — it’s something I’m very dissatisfied with. Also, Dario doesn’t like Sam Altman; in fact, the reason the Anthropic team left OpenAI was because they didn’t trust Sam Altman, otherwise they'd still be at OpenAI.
Musk
I think Dario is a very principled person; he cares a lot, and cares about the future of the world. So far, everyone I’ve met at Anthropic is the same; no one has struck me as malicious — they all seem well-intentioned. The road to hell is often paved with good intentions, though; a few are well-intentioned, but we shouldn’t be complacent. Still, if something needs to be discussed, let’s discuss it. For the good of the world, let’s put aside personal differences.
Risk Warning and DisclaimerThe market has risks, and investment needs caution. This article does not constitute personal investment advice and does not take into account individual users’ investment objectives, financial situations, or needs. Users should consider whether any opinions, views, or conclusions in this article suit their particular circumstances. Invest accordingly at your own risk. ```