Bernie Sanders and Steve Bannon agree on AI at ‘pro-human’ gathering

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

Fears about the rapid advancement of artificial intelligence have bent the American political spectrum to the point that Sen. Bernie Sanders, the democratic socialist from Vermont, appeared on the same stage as Steve Bannon, the former Trump adviser and MAGA fundamentalist, at an event in Washington, DC, on Tuesday.

While the two did not appear onstage together, they spoke just moments apart at the Pro-Human Assembly to call for urgent action to rein in AI.

Both men see what Bannon called a “Cold War moment.” And both raised concerns about the benefits of AI being concentrated among a small number of tech billionaires, while the American public is left to grapple with its risks and pitfalls.

The event came as debate heats up around how to handle the serious risks presented by AI, which some experts and leaders believe could be existential, but which President Donald Trump has called a “hoax.” Rogue AI hacking incidents and AI companies’ own reports about misuse of their tools have added to the fears.

Sanders raised concerns that AI will wipe out jobs, create a “mass surveillance state,” compromise people’s well-being and jeopardize the integrity of elections.

Bannon similarly warned about the impact of AI on an American public that largely did not ask for it. He said Americans “are not going to be supplicants” to AI or AI companies, whom he accused of trying to form a cartel to control the technology.

Both Bannon and Sanders want to put the brakes on the development of superintelligence — an agreement that is notable on its own. But it was clear from their remarks that they diverge on how to move forward, illustrating a split that exists all around Washington.

Sanders has proposed legislation that would give the US government a direct stake in AI companies through a one-time 50% tax on their value to create an American sovereign wealth fund and a law to ban superintelligence that could subvert human control.

Bannon, by contrast, said laws take too much time. He wants President Donald Trump to use executive action to gin up some kind of regulatory agency, although he shared no specifics on what that could look like.

Sanders talked about working with other countries because this is “humanity’s problem,” encouraging Trump to draw up a treaty with China when he meets with Chinese President Xi Jinping later this month. But Bannon said a treaty would take too long and give dangerous power to China; he called instead for barring Chinese nationals from US universities and labs.

Finn Gomez/Getty Images

Comparing the AI race to the nuclear arms race, Sanders pointed to US President Ronald Reagan and Soviet Premier Mikhail Gorbachev putting aside differences to form a nuclear disarmament treaty.

“With the survival of humanity at stake, they found a way to sign that treaty,” Sanders said.

Moments later, Bannon gave a very different history lesson.

“Our father and grandfathers,” he said, “stood up to the Soviet Union, and the first thing in standing up to them was to cut them off.”

There should be no deals or agreements with China, Bannon said.

Sanders, who did not mention Bannon, argued that Trump’s ignorance of AI issues is embarrassing. Bannon acknowledged that he and Sanders have very different views of Trump. “We’re trying to raise above partisanship, and we have to,” he said.

While the two advocated for very different approaches to controlling AI, their level of agreement is notable.

Neither Sanders nor Bannon has any faith in AI companies’ CEOs. Sanders demanded that they stop pouring money into US elections; Bannon said they cannot be trusted.

Organizers with the Future of Life Institute said their “pro-human declaration” — which calls for measures such as an “off switch” for powerful AI systems — now has 1 million signatories, including the American Federation of Teachers, SAG-AFTRA and the online safety group Heat Initiative.

The larger alignment between Sanders and Bannon shows how political lines are being redrawn over how to handle the controversial issue.

At the heart of the debate is the question of whether AI development must be slowed to prevent catastrophic risks or if any delay could put the United States behind China in the race to highly powerful AI and blunt the technology’s economic benefits.

A chorus of top AI leaders have in recent days called for government and the corporate sector to implement additional guardrails and intentionally slow the pace of development to keep it from getting out of hand. Still, others in the industry question whether such calls stem, at least in part, from a desire to shape regulation in a way that could be friendly to certain companies.

Anna Moneymaker/Getty Images

Trump has called fears around AI a “hoax” in a call with Nvidia CEO Jensen Huang during a live event on Monday. He has rejected the idea of additional AI controls, saying he wants the US to stay ahead of China . House Speaker Mike Johnson has also said that Congress shouldn’t rush to act. That’s despite a growing anti-AI push on the right , including from Bannon, who for months raised concerns about AI’s benefits to what he calls the tech “oligarchy.”

Democrats ranging from House Minority Leader Hakeem Jeffries to former President Barack Obama have called for swift action to rein in AI.

A June report from Pew Research Center found that 40% of US adults believe AI will be negative for society over the next 20 years, compared with just 16% who believe it will be positive and 31% who said it would be equally positive and negative.

The issue does not fall along traditional partisan lines. One thing to watch will be whether Republicans take the cue from Trump that efforts to rein in AI are part of a “hoax.”

Tuesday’s event was scheduled to include a wide range of speakers addressing how to shape AI to benefit humanity, including actor Joseph Gordon-Levitt, who has raised concerns about AI’s impact on Hollywood and children, and American Federation of Teachers President Randi Weingarten, who has sought to develop recommendations and trainings for the use of AI in schools. One panel also included the parents of children who died by suicide after interacting with AI chatbots, a stark reminder that much of humans’ interaction with AI occurs behind closed doors and can already have disastrous consequences.

“When you are racing towards a cliff you don’t just ease up on the gas pedal. You hit the brakes,” Sanders said.

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。

对人工智能快速发展的担忧已经扭曲了美国的政治格局，以至于佛蒙特州的民主社会主义参议员伯尼·桑德斯周二在华盛顿特区的一个活动上与前特朗普顾问、MAGA 原教旨主义者史蒂夫·班农同台亮相。

虽然两人没有同时出现在舞台上，但他们在“支持人类大会”上相隔几分钟先后发表了讲话，呼吁采取紧急行动来遏制人工智能。

两人都看到了班农所说的“冷战时刻”。两人都对人工智能的益处集中在少数科技亿万富翁手中，而美国公众却不得不承受其风险和弊端表示担忧。

此次活动正值围绕如何应对人工智能带来的严重风险的辩论愈演愈烈之际。一些专家和领导人认为，人工智能可能构成生存威胁，但美国总统唐纳德·特朗普却称其为“骗局”。人工智能黑客攻击事件以及人工智能公司自身关于其工具被滥用的报告，都加剧了人们的担忧。

桑德斯担心人工智能会消灭工作岗位，建立“大规模监控国家”，损害人们的福祉，并危及选举的公正性。

班农也同样警告说，人工智能会对美国公众产生影响，而美国公众大多并未要求这项技术。他说，美国人“不会向人工智能或人工智能公司卑躬屈膝”，他还指责这些公司试图组成卡特尔来控制这项技术。

班农和桑德斯都希望遏制超级智能的发展——这一点本身就值得关注。但从他们的言论中可以明显看出，他们在如何推进超级智能发展的问题上存在分歧，这也反映了华盛顿政坛普遍存在的这种分裂局面。

桑德斯提出了一项立法提案，该提案将通过对人工智能公司价值一次性征收 50% 的税，使美国政府直接持有这些公司的股份，从而创建一个美国主权财富基金；他还提出了一项法律，禁止可能颠覆人类控制的超级智能。

相比之下，班农则认为立法过程耗时太长。他希望特朗普总统通过行政命令迅速建立某种监管机构，但他并未透露具体细节。

桑德斯谈到要与其他国家合作，因为这是“全人类的问题”，并鼓励特朗普在本月晚些时候与中国国家主席习近平会晤时与中国起草一项条约。但班农表示，签署条约耗时太长，而且会赋予中国危险的权力；他呼吁禁止中国公民进入美国大学和实验室。

Finn Gomez/Getty Images

桑德斯将人工智能竞赛比作核军备竞赛，并指出美国总统罗纳德·里根和苏联领导人米哈伊尔·戈尔巴乔夫搁置分歧，缔结了核裁军条约。

桑德斯说：“在人类生存受到威胁的情况下，他们想方设法签署了这项条约。”

片刻之后，班农又讲述了一个截然不同的历史故事。

他说：“我们的父亲和祖父们曾挺身对抗苏联，而对抗苏联的第一步就是切断与他们的关系。”

班农表示，不应该与中国达成任何交易或协议。

桑德斯没有提及班农，而是指出特朗普对人工智能问题的无知令人尴尬。班农承认他和桑德斯对特朗普的看法截然不同。“我们试图超越党派之争，而且我们必须这样做，”他说。

虽然两人在控制人工智能方面提倡截然不同的方法，但他们的共识程度却很显著。

桑德斯和班农都对人工智能公司的首席执行官们缺乏信任。桑德斯要求他们停止向美国大选投入资金；班农则表示他们不可信赖。

未来生命研究所的组织者表示，他们的“亲人类宣言”（呼吁采取措施，例如为强大的人工智能系统设置“关闭开关”）目前已有 100 万人签署，其中包括美国教师联合会、美国演员工会-美国电视和广播艺术家联合会以及网络安全组织 Heat Initiative。

桑德斯和班农之间更广泛的结盟表明，在如何处理这一争议性问题上，政治界限正在重新划分。

这场辩论的核心问题是：是否必须放慢人工智能的发展速度以防止灾难性风险，或者任何延误是否会导致美国在高度强大的人工智能竞赛中落后于中国，并削弱该技术的经济效益。

近日，众多人工智能领域领军人物呼吁政府和企业界加强监管，有意放慢人工智能发展速度，以防止其失控。然而，业内也有人质疑，这些呼吁是否至少部分源于某种意图，即希望通过制定有利于某些公司的监管政策来规避风险。

安娜·莫尼梅克/盖蒂图片社

周一，特朗​​普在与英伟达首席执行官黄仁勋的通话中，称人们对人工智能的担忧是“骗局”。他拒绝了加强人工智能管控的想法，称希望美国保持对中国的领先地位。众议院议长迈克·约翰逊也表示，国会不应仓促采取行动。尽管右翼人士，包括班农在内，对人工智能的反人工智能呼声日益高涨。班农几个月来一直担忧人工智能会给所谓的科技“寡头”带来好处。

从众议院少数党领袖哈基姆·杰弗里斯到前总统巴拉克·奥巴马，众多民主党人都呼吁迅速采取行动来限制人工智能的发展。

皮尤研究中心 6 月份的一份报告显示，40% 的美国成年人认为人工智能在未来 20 年内会对社会产生负面影响，而只有 16% 的人认为它会带来正面影响，31% 的人则认为它会带来正面和负面的双重影响。

这个问题并非传统意义上的党派之争。值得关注的一点是，共和党人是否会效仿特朗普，认为限制人工智能的努力是一场“骗局”。

周二的活动原计划邀请众多演讲嘉宾，探讨如何引导人工智能造福人类。其中包括曾对人工智能对好莱坞和儿童的影响表示担忧的演员约瑟夫·戈登-莱维特，以及致力于制定人工智能在学校应用建议和培训的美国教师联合会主席兰迪·温加滕。其中一个小组讨论环节还邀请了因与人工智能聊天机器人互动而自杀身亡的儿童的家长，这令人警醒地提醒人们，人类与人工智能的互动大多发生在私下，而这些互动可能已经造成灾难性的后果。

“当你驾车冲向悬崖时，你不会只是松开油门，而是会踩刹车，”桑德斯说。