Trump really wants to make ‘super intelligence’ happen. Is it going to happen?

Kevin Dietsch/Getty Images

A version of this story appeared in CNN’s What Matters newsletter. To get it in your inbox, sign up for free here .

Unable to convince Americans to embrace the AI data centers they hate , President Donald Trump is instead exercising the one power he believes is absolute in the presidency: a rebrand.

But unlike Lake America , the Gulf of America, military bases or a mountain, this change is about altering public opinion rather than scoring political points.

Trump first told fellow world leaders at the United Nations last month that “use of the word artificial makes intelligence fake, it makes it sound fake.”

After Trump signed an executive order in late September, the US government has been changing official documents and websites. Cabinet secretaries and other aides immediately used the new term.

Elon Musk said on X that he would change the name of his company from SpaceXAI to SpaceXSI, although that particular change has not yet occurred.

Trump has refused to answer questions that use the old term. He told supporters at a campaign rally in Nebraska that he would henceforth repurpose “artificial” — an “ugly” word, he said — for the news media.

On Thursday, the president said on social media that anyone who uses the term artificial intelligence instead of super intelligence will be viewed by his White House as “THE ENEMY!”

All of this might feel very much like Trump’s “weave” — the mental meanderings of Trump’s public statements — but it has also been fascinating to watch the infrastructure of the federal government respond to what many Americans might write off as a semantic distinction.

There are things that are out of Trump’s power — passing spending bills, renaming the Kennedy Center, ending his war against Iran — but he has shown that presidents have power to change names.

It’s also evidence he knows the unpopularity of data centers and concerns about AI in general at a time when he is a major booster of AI.

Meta CEO Mark Zuckerberg has long pushed to view AI as superintelligence. In August, he published a 6,000-word treatise about how Meta would approach AI. In rose assessments of the future, he referred to the promise of superintelligence. When he referred to AI, it was usually in reference to what he argued were unfounded worries. Zuckerberg said Meta will be committed to an open source model that provides for superintelligent agents working on behalf of every individuals.

OpenAI CEO Sam Altman told Politico’s “Decoded” podcast there are no plans to change the name of his company to OpenSI. But he did say that people in the industry have been using the term superintelligence for a while.

Brendan McDermid/Reuters

It might be a better term, ultimately, than artificial intelligence, but “still sort of understates the magnitude of what I think is happening,” Altman said.

Altman predicted people would still use the term AI and questioned whether Trump’s rebrand would accomplish its goal of changing people’s opinions of AI.

“Does superintelligence sound obviously less scary to people than artificial intelligence?” Altman wondered. Time will tell.

What is definitely true is that the concept of superintelligence, while it sounds better in Trump’s view, is arguably distinct from AI and should be no less intimidating to Americans.

The term artificial intelligence has its roots in the 1950s, when a group of Dartmouth professors held a summertime conference “on the basis of the conjecture that every aspect of learning or any other feature of intelligence can in principle be so precisely described that a machine can be made to simulate it.”

Superintelligence, at least academically, seems more like the next step of artificial intelligence, or artificial generative intelligence. That happens when the machines learn.

The Oxford philosopher Nick Bostrom has written extensively about superintelligence, including in the 2014 book “Superintelligence: Paths, Dangers, Strategies.”

Here’s how Bostrum first defined superintelligence in a 1997 essay:

Asked how he views Trump’s rebrand, Bostrum said that from a technical perspective AI has traditionally “been this wide umbrella term that refers to every form of intelligent information processing in machines.”

We’re talking everything from the Google searches and email spam filters, the current crop of AIs and “hypothetical future more powerful ones.”

“If the idea is to use the new word superintelligence to cover all of that, it’s a little bit of a stretch, maybe, to call my spam filter a superintelligence,” Bostrum told CNN.

In the theoretical evolution of AI, what comes next, he said is AGI, or artificial general intelligence, which refers to a system that has the same kind of learning and reasoning as humans, something Atman says might already have occurred.

“Superintelligence was reserved for this sort of hypothetical future form of artificial intelligence that not only matched but radically exceeded humans across all cognitive domain,” according to Bostrum.

That is exactly the thing that frightens the AI “doomers.” But it is incomprehensible to Trump that the machines could outwit him.

“The only control or ‘guardrails’ that AI needs is a STRONG AND SMART (High IQ!) PRESIDENT, and the U.S.A. has that, in spades!” Trump wrote in September, arguing against any sort of AI regulation.

Anna Moneymaker/Getty Images

Bostrum has written both about potential pitfalls of superintelligence as well as the enormous potential to improve and extend human life. But by changing the name, Trump seems to just want to control the narrative.

Bostrum doesn’t necessarily think it’s a bad idea.

“It’s not up to me to decide how English language should be used,” he said. “I mean, it’s good for my book sales, I guess.”

He does see the rhetorical argument for a new term: It’s harder to argue against super intelligence than artificial intelligence.

“If you define superintelligence as we already have superintelligence and it hasn’t caused any catastrophe yet, then what’s the big deal? And you then would need to introduce some new term to refer to this like hypothetical future capability that’s even more powerful.”

Trump knows a thing or two about saying things so much that they take hold in the public. He rarely speaks in public without repeating the same talking points about “fake” news, the “radical left” Democrats, or about how wonderful his own leadership has been, regardless of what the facts are.

At the UN, Trump compared his rebrand to the evolution of “global warming,” which he said was “reborn to climate change because the planet was cooling, not warming, and nobody was dead.”

That’s not exactly what happened, as anyone paying attention to the warming climate in recent years can pretty easily confirm.

Kevin Dietsch/Getty Images

本文的另一个版本曾刊登在 CNN 的“要闻速递”（What Matters）新闻简报中。如需订阅，请点击此处免费注册。

由于无法说服美国民众接受他们所厌恶的人工智能数据中心，唐纳德·特朗普总统转而行使他认为总统职位上唯一绝对的权力：重新打造品牌。

但与美洲湖、美洲湾、军事基地或山脉不同，这种变化是为了改变公众舆论，而不是为了获得政治利益。

特朗普上个月在联合国首次告诉其他国家领导人，“使用‘人工智能’这个词会让智能听起来像是假的，让智能听起来像是假的。”

特朗普9月下旬签署行政命令后，美国政府开始更改官方文件和网站。内阁部长和其他助手立即开始使用新术语。

埃隆·马斯克在 X 频道上表示，他将把公司名称从 SpaceXAI 改为 SpaceXSI，尽管这一变更尚未发生。

特朗普拒绝回答使用旧词的问题。他在内布拉斯加州的一次竞选集会上告诉支持者，他今后将对“人工的”（他称之为“难听的”词）一词进行重新定义，供新闻媒体使用。

周四，总统在社交媒体上表示，任何使用“人工智能”而不是“超级智能”的人都将被白宫视为“敌人”！

这一切可能感觉很像特朗普的“编织”——特朗普公开声明中漫无边际的思维游移——但观察联邦政府的基础设施如何应对许多美国人可能认为只是语义上的区别，也令人着迷。

有些事情超出了特朗普的权力范围——比如通过支出法案、更改肯尼迪中心的名称、结束对伊朗的战争——但他已经表明，总统有权更改名称。

这也证明，尽管他本人是人工智能的主要支持者，但他也知道数据中心不受欢迎，并且对人工智能普遍存在担忧。

Meta首席执行官马克·扎克伯格长期以来一直倡导将人工智能视为超级智能。今年8月，他发表了一篇长达6000字的文章，阐述了Meta将如何开发人工智能。在对未来进行评估时，他多次提及超级智能的前景。而当他提到人工智能时，通常指的是他认为毫无根据的担忧。扎克伯格表示，Meta将致力于采用开源模式，为每个人提供超级智能体。

OpenAI首席执行官萨姆·奥特曼在接受Politico的“Decoded”播客采访时表示，公司目前没有计划更名为OpenSI。但他确实提到，业内人士使用“超级智能”这个词已经有一段时间了。

布伦丹·麦克德米德/路透社

奥特曼说，从长远来看，它可能比人工智能更合适，但“仍然有点低估了我认为正在发生的事情的规模”。

奥特曼预测人们仍会使用人工智能一词，并质疑特朗普的品牌重塑能否实现改变人们对人工智能的看法这一目标。

“超级智能听起来是不是比人工智能更容易让人接受？”奥特曼心想。时间会证明一切。

可以肯定的是，超级智能的概念虽然在特朗普看来听起来更好，但它与人工智能显然有所不同，而且对美国人来说，超级智能同样令人感到恐惧。

“人工智能”一词起源于 20 世纪 50 年代，当时一群达特茅斯学院的教授举行了一次夏季会议，“基于这样的猜想：学习的每个方面或智能的任何其他特征原则上都可以被精确描述，从而制造出一台机器来模拟它。”

至少从学术角度来看，超级智能更像是人工智能或人工生成智能的下一个发展阶段。而人工生成智能的出现，正是机器学习的过程。

牛津大学哲学家尼克·博斯特罗姆曾就超级智能撰写过大量文章，其中包括 2014 年出版的《超级智能：路径、危险、策略》一书。

以下是博斯特鲁姆在1997年的一篇文章中首次对超级智能的定义：

当被问及如何看待特朗普的品牌重塑时，博斯特鲁姆表示，从技术角度来看，人工智能传统上“是一个涵盖机器中各种形式的智能信息处理的广泛术语”。

我们谈论的话题涵盖了从谷歌搜索和电子邮件垃圾邮件过滤器，到当前的人工智能以及“假设的未来更强大的人工智能”。

博斯特鲁姆告诉 CNN：“如果用‘超级智能’这个新词来涵盖所有这些，那么把我的垃圾邮件过滤器称为超级智能可能有点牵强。”

他表示，在人工智能的理论发展过程中，下一步将是通用人工智能（AGI），指的是具有与人类相同学习和推理能力的系统，阿特曼认为这种情况可能已经发生了。

博斯特鲁姆表示：“超级智能指的是这种假想的未来人工智能形式，它不仅能与人类匹敌，而且在所有认知领域都远远超越了人类。”

这正是人工智能“末日论者”们所恐惧的。但特朗普无法理解机器竟然能智胜他。

“人工智能唯一需要的控制或‘护栏’就是一位强大而聪明（高智商！）的总统，而美国在这方面人才济济！”特朗普在9月份写道，反对任何形式的人工智能监管。

安娜·莫尼梅克/盖蒂图片社

博斯特伦曾撰文探讨超级智能可能带来的潜在隐患，以及它在改善和延长人类寿命方面的巨大潜力。但特朗普更改名称，似乎只是想控制舆论。

博斯特鲁姆并不认为这是一个坏主意。

“我无权决定英语应该如何使用，”他说。“我的意思是，我想这对我的书销量有好处。”

他认为使用新术语的修辞论点是：反对超级智能比反对人工智能更难。

“如果你把超级智能定义为我们已经拥有超级智能，而且它还没有造成任何灾难，那这有什么大不了的呢？这样一来，你就需要引入一些新的术语来指代这种假设的未来能力，比如更强大的能力。”

特朗普深谙如何通过反复强调某些观点来使其深入人心。他很少在公开场合不重复那些老生常谈的论调，比如“假新闻”、“激进左翼”民主党人，或者吹嘘自己的领导有多么出色，而全然不顾事实真相。

在联合国，特朗普将他的品牌重塑比作“全球变暖”的演变，他说“全球变暖”之所以“演变为气候变化”，是因为地球当时正在变冷，而不是变暖，而且也没有人因此丧命。

事实并非如此，任何关注近年来气候变暖的人都可以很容易地证实这一点。