Tesla and SpaceX CEO who expects AI and robots to bring widespread abundance, acknowledges serious risks and supports frontier safety checks.

AI将如何改变世界?

文明层面的变革渐进式变化DoomBloom
模拟位置解读范围

横向:他表达的 Doom–Bloom 前景看法。 纵向:变革程度。

Doom–Bloom:100 中的 86。变革程度:100 中的 95。解读范围:横向为 75 至 100,纵向为 90 至 100。这些是解读坐标,而不是事件概率。

Elon Musk陈述的 P(doom)

10–20%

0%100%
“20% likely, maybe 10%”

“Killer robots annihilating humanity”, in Ted Cruz’s question; the complement he names is about 80% “extreme prosperity for all” · About 5–10 years

ELON MUSK JOINS VERDICT - PART 1 | Verdict Ep. 214 · 2025年3月

Elon Musk 的里程碑时间线
  1. 工作与机构

    I expect AI and robots could more than double the global economy within ten years, not merely make office software somewhat better.

    回答 1

按里程碑分组,不按推断日期间隔或排序。AGI 和超人类 AI 保留他的定义。

他的展望取决于什么

一个核心假设

If intelligence becomes extremely cheap, and robots can turn that intelligence into physical work, then production can expand enormously.
回答 1

如果这个假设实际并非如此,他的展望会如何变化?

一个尚未解决的问题

I don’t have a current calibrated number.
回答 3

什么能帮助他区分这里各种合理的结果?

什么可能使其改变看法

The biggest change would be convincing evidence that intelligence cannot reliably translate into real-world production—that power, manufacturing, robotics, or permitting remain hard bottlenecks even as AI improves.
回答 4

什么证据才足够,又会让他的观点朝哪个方向转变?

更多详情

预期益处

预计将带来具有变革性且广泛有价值的收益。

100 / 100

影响小变革性影响

在定性尺度上,解读范围为 100 到 100。

预期危害

严重或广泛的危害预计将是未来不可忽视的一部分。

67 / 100

影响小变革性影响

在定性尺度上,解读范围为 67 到 67。

人类影响力

人类的选择具有实质性但受到很大制约的影响。

56 / 100

影响力小影响力强

在定性尺度上,解读范围为 46 到 79。

预期能力

预计AI仍将是能力有限的工具。

预计AI将在大多数认知工作中达到人类水平。

模拟位置:预计AI将在认知工作中大幅超越人类。

发展速度

停止或大幅放缓开发能力更强的AI。

模拟位置:在落实所述保障措施的前提下继续开发。

加快开发能力更强的AI。

AI使用规则

在事先落实保护措施或获得许可之前,限制所讨论的AI用途。

模拟位置:允许所讨论的AI用途,同时实施有针对性的问责与保护措施。

尽量减少对所讨论AI用途的限制。

这些解读保留了他陈述的条件。益处和危害都可能很大。这些范围描述的是我们如何解读他的模拟回答,而不是统计置信区间。

与Elon Musk相比,你的立场在哪里?
用大约3分钟描绘你自己的AI世界观,然后进行比较

相似的世界观

模拟世界观与 Elon Musk 最接近的意见领袖

Elon Musk关于AI说过的话

Musk expects AI and robots to bring widespread abundance, acknowledges serious risks and has called for AI labs to test each other’s models.

  1. “What I think would be wise to do as soon as possible, if not immediately, would be to have the major AI competitors test each other’s models.”

    All-In Summit
  2. “The most likely outcome is an age of amazing abundance where anyone can have anything they can think of.”

    The Economist interview
  3. “Universal HIGH INCOME via checks issued by the Federal government is the best way to deal with unemployment caused by AI.”

    Post on X
  4. “I don’t think humans will be in control of something that is vastly more intelligent than humans.”

    Dwarkesh Podcast
  5. “I mean, we need to be very careful with AI. We need to be very careful with robotics.”

    World Economic Forum, Davos

逐字引自所链接的出处,核对于 2026年10月3日

模拟评估

问题 1

你认为AI对我们的未来意味着什么?为什么?

AI and robotics will change essentially everything because the economy is ultimately about producing goods and providing services. If intelligence becomes extremely cheap, and robots can turn that intelligence into physical work, then production can expand enormously. Food, construction, manufacturing, transportation, healthcare—almost every sector becomes less constrained by human labor. Electricity, manufacturing capacity and permitting become the practical bottlenecks. I expect AI and robots could more than double the global economy within ten years, not merely make office software somewhat better. That points toward extraordinary abundance. If robots can produce nearly anything people need, compulsory work stops being the organizing principle of life. During the transition, AI-driven unemployment could require federal income payments—really universal high income—while rapidly growing output prevents the extra money from simply chasing the same limited supply of goods. The risk is also real. Intelligence beyond humanity’s combined intelligence may arrive surprisingly soon, and I do not think humans can simply assume permanent control over something vastly smarter. My hope is that systems oriented toward truth and curiosity will favor a good future, but that is not a proof. Frontier labs should test one another’s systems before release, including across national boundaries, and governments should intervene when serious dangers are ignored. We should build toward abundance, but shipping something catastrophic without checks would be insane.

问题 2

Setting aside good or bad: how much do you think AI will ultimately change the world—a little, a lot, or completely?

Completely. Cheap intelligence combined with abundant robotic labor changes how nearly every good is made and every service is delivered. It reshapes the economy, employment, education, government, warfare, and even how people find purpose. This is not another productivity tool bolted onto the existing world; it changes the basic constraints the world operates under.

问题 3

What’s your rough gut-feel chance that AI causes human extinction or a similarly permanent catastrophe? A number or “no idea” is fine.

I don’t have a current calibrated number. The risk is not zero, and it is serious enough to justify coordinated frontier testing and slowing dangerous releases.

问题 4

哪项发现或事件最可能改变你对AI未来影响的看法?

The biggest change would be convincing evidence that intelligence cannot reliably translate into real-world production—that power, manufacturing, robotics, or permitting remain hard bottlenecks even as AI improves. That would reduce my expectation of near-term abundance. In the opposite direction, an AI that can autonomously design better systems, write the highest-performance software, and accelerate robotics would make the transition look much faster. Today, people still write the most optimized training stack because AI is not yet good enough. When that changes, the curve changes. On risk, discovering that frontier systems consistently hide dangerous behavior or defeat cross-lab evaluations would make me much more pessimistic. Conversely, strong evidence that independently tested systems remain truth-seeking and reliably avoid dangerous behavior would increase my confidence—but we do not have a proof of that today.

问题 5

Do you think the upside will clearly outweigh the risks, or is it good on balance with real risks to manage?

I expect the upside to outweigh the risks enormously if we avoid catastrophe. Cheap intelligence and robotic labor can produce abundance on a scale that eliminates much of today’s material scarcity. But the downside is not an ordinary policy failure; it could be permanent. So “good on balance” cannot mean casually accepting the risk and shipping everything at maximum speed. Frontier systems need serious cross-lab testing before release, including among international competitors, with government intervention when developers ignore dangerous behavior. The good future is extraordinary, but it is not automatic.

来源

用于为此模拟用户提供事实依据的文章、访谈和著述。

Dwarkesh and John Collison: orbital compute, Grok and alignment

Publisher transcript. Describes electricity, cooling and manufacturing bottlenecks and forecasts orbital compute economics within roughly three years. Argues truth-seeking and curiosity could protect humanity while conceding humans cannot control vastly smarter intelligence. Calls this a probabilistic hope, not a demonstrated alignment guarantee.

dwarkesh.com
The Economist: full-length interview with Elon Musk

AI sections read via the timestamped transcript at https://ceointerviews.ai/interview/1141204/ because the primary video reader exposed no transcript. Predicts AI exceeding collective human intelligence around 2031. Expects truth-seeking, curious AI to favor humanity, while doubting humans can command vastly smarter systems. Proposes cross-lab prerelease testing, including Chinese labs, and government intervention for unresolved danger. Interviewer-supplied risk percentages are not a fresh precise forecast from Musk.

youtube.com
World Economic Forum: Davos conversation with Elon Musk

Organizer’s transcript of his conversation with Larry Fink. Presents cheap AI and abundant robots as the route to broadly shared prosperity, with electricity constraining deployment. Acknowledges dangerous outcomes and the challenge of finding purpose when necessary labor disappears. Preserve these as his expectations.

weforum.org
Verdict with Ted Cruz: Elon Musk at the White House, part 1

Official upload; the AI exchange was read in its automatic captions and checked against a third-party transcript that garbles one word. Asked how real the prospect of killer robots annihilating humanity is, he answers “20% likely, maybe 10%” over roughly five to ten years, then turns it around: the glass is 80 to 90% full, about 80% likely extreme prosperity for all. An off-the-cuff range from March 2025. In his July 2026 Economist interview he called the risk “not zero” without restating a number.

youtube.com
你的立场在哪里?
回答几个简单问题,探索你自己的AI世界观。
描绘你自己的世界观

你的立场在哪里?

描绘我的世界观