精读《经济学人》：AI 该不该踩刹车？华盛顿已经吵翻了

这篇《经济学人》文章聚焦华盛顿的 AI 监管之争：一边是科学家警告 AI 可能"杀死全人类"，一边是白宫高调反对"扼杀行业"，两党、科技公司与各州政府各有算盘。本文带你读原文、学词汇、练表达。

01 文章摘要

事件核心：9 月 15 日华盛顿举办"人类利益大会"，桑德斯与班农这两位立场迥异的政治人物罕见同台，呼吁立法为 AI 降温。

各方立场：特朗普称 AI 风险是"骗局"，主张全力追赶中国；民主党领袖呼吁紧急立法，共和党议长则警告"仓促监管会输掉对华竞赛"。

民意与现状：民调显示约八成美国人支持监管，即使这会拖慢创新；目前立法主要停留在州层面（加州、纽约已先行立法），联邦层面提案混杂、分歧明显。

产业态度与趋势：Anthropic 最积极支持监管，但没有任何一家实验室全力支持某项大提案；随着中期选举与总统竞选推进，两党立场可能进一步强硬。

02 原文精读（节选）

节选一：Both men warned of the unchecked power of tech "oligarchs" and called for urgent legislation to slow down the technology.两人都警告科技"寡头"的权力不受约束，并呼吁紧急立法为 AI 踩下刹车。unchecked 意为"不受约束的"，词根 check 有"控制、制止"之义。

节选二：Mr Coxon said that he and his colleagues believed that AI could "kill us all" by the end of the decade, a message that he promptly delivered across prime-time television.考克森表示，他和同事相信 AI 可能在本十年末"杀死我们所有人"，并迅速把这一警告带上黄金时段电视节目。a message that... 是同位语加定语从句，对整句内容作补充说明。

节选三：Were it not for opposition from the likes of Leading the Future, a super PAC..., both laws would have been bolder.若非"引领未来"等组织的反对，这两部法律本会更激进。Were it not for 是 if it were not for 的倒装虚拟语气，意为"若非、要不是"。

节选四：Voters' patience for "sensible" regulation may wear thin.选民对"明智监管"的耐心可能逐渐耗尽。wear thin 原指布料"越穿越薄"，引申为"逐渐失效、令人失去耐心"。

03 重点单词

• firebrand n. 煽动者、激进分子｜political firebrands from opposite ends of the spectrum

• unchecked adj. 不受约束的｜the unchecked power of tech "oligarchs"

• oligarch n. 寡头｜tech oligarchs

• hoax n. 骗局、谎言｜AI's supposed risks were a "hoax"

• extinction n. 灭绝｜put the probability of extinction at more than 10%

• partisan adj. 党派的｜a partisan divide remains

• hasty adj. 仓促的｜warned against hasty regulation

• liability n. 法律责任｜exposing firms to more liability

• rogue adj. 失控的｜OpenAI's agents went rogue

• existential adj. 关乎生存的｜cover only existential risks

• catastrophic adj. 灾难性的｜posing imminent "catastrophic" risk

• hodgepodge n. 大杂烩｜a hodgepodge of proposals

04 重点短语与句式

地道短语

• from opposite ends of the spectrum：来自政治光谱的两端（立场迥异）

• in common cause：为共同目标而携手

• sit up：警觉起来、认真对待

• on pain of ... penalties：违者将面临……的罚款

• water down：削弱、淡化

• go rogue：失控、不受管束

• wear thin：逐渐失效、令人失去耐心

• ramp up：升级、加剧

句式精讲

• Were it not for opposition..., both laws would have been bolder. 若非……，这两部法律本会更激进——if it were not for 的倒装虚拟语气，常用于书面语。

• The trickier question is what that regulation should be. 更难的问题是监管应当如何——what 引导表语从句。

• Eight in ten say they support regulation, even if it slows down innovation. 十人中有八人支持监管，即使这会拖慢创新——"数据 + even if 让步"是外刊高频表达。

05 背景小知识

• 《经济学人》（The Economist）：1843 年创刊的英国周刊，以深度分析、匿名写作与犀利文风著称，是英语精读的经典素材。

• Anthropic 与 OpenAI：美国两大头部 AI 实验室，文中提到的离职研究员考克森即出自 Anthropic。

• Super PAC：超级政治行动委员会，可无限额筹集竞选资金，是影响美国政策的重要力量。

• Truth Social：特朗普创办的社交平台，文中"AI 风险是骗局"等表态即发布于此。

• 中期选举：美国总统任期中期举行的国会选举，常被视作民意风向标。

06 思考题（用英语回答）

1. Should governments slow down AI development, even if it slows innovation?

2. Is "self-policing" by AI companies enough to protect the public?

3. What might make AI a truly "existential risk" to humanity?

07 学习建议

• 第一步：先读"文章摘要"把握大意，再精读原文节选，降低理解压力。

• 第二步：把重点单词放回原句记忆，用"摘抄原句 + 自造新句"的方式巩固。

• 第三步：模仿"句式精讲"里的结构各仿写 1 句，练习主动输出。

• 第四步：每周精读 1 篇同类报道，积累专有名词与观点表达。

觉得有用的话，欢迎点赞、在看、转发；关注我们。

原文阅读：

The debate over AI has taken over Washington

Whether it will act is another matter

FEW EVENTS have brought together Bernie Sanders and Steve Bannon, political firebrands from opposite ends of the spectrum, in common cause. But the Pro-Human Assembly in Washington, DC on September 15th, a gathering of people alarmed by the rapid advance of artificial intelligence, managed it. Both men warned of the unchecked power of tech “oligarchs” and called for urgent legislation to slow down the technology.

Donald Trump, the president, is unmoved by the torrent of such demands. AI’s supposed risks to humanity, he wrote on social media, were a “hoax”. America’s aim was to beat China; proponents of a slowdown were “treasonists”. But in recent days, political momentum for tighter controls has gathered with unusual speed.

The Pro-Human Assembly and Mr Trump’s flurry of posts on the subject came less than a week after the public resignation of Jacob Coxon, a researcher at Anthropic, an AI firm. Mr Coxon said that he and his colleagues believed that AI could “kill us all” by the end of the decade, a message that he promptly delivered across prime-time television. Another top scientist at the firm put the probability of extinction at more than 10%.

That has made many politicians sit up. But despite the mixed politics of the Pro-Human Assembly, a partisan divide remains. Hakeem Jeffries and Chuck Schumer, the Democratic leaders in the House and Senate, respectively, called for urgent action. Mike Johnson, the Republican speaker of the House, warned against hasty regulation. “If Congress just races in and does some sort of emergency session to try to regulate AI,” said Mr Johnson, “we will lose the race to China”. AI labs, he argued, should “self-police”.

In general, polls show that Americans disagree: eight in ten say they support regulation, even if it slows down innovation. The trickier question is what that regulation should be. To date most activity has been in states. Last September Gavin Newsom, California’s governor, signed a law requiring AI firms to report safety incidents within 15 days, on pain of up to $1m in penalties. Three months later Kathy Hochul, New York’s governor, signed a law demanding faster disclosure and imposing larger fines. Both laws would have been bolder, with higher penalties and exposing firms to more liability, were it not for opposition from the likes of Leading the Future, a super PAC whose supporters include Greg Brockman, a founder of OpenAI, another firm. Clear gaps in state laws remain. California’s law did not apply to the now-notorious Hugging Face incident, when OpenAI’s agents went rogue, because it occurred in an “evaluation phase” designed to elicit bad behaviour.

If proponents of federal regulation want to secure bipartisan support, they may have to compromise in two areas, argues Brad Carson, the co-founder of Public First Action, which advocates for AI regulation and is backed by Anthropic. One is how easy it is for the government to shut down an errant model. A second is whether bills should cover only existential risks, such as AI models’ ability to create biological weapons, or also include issues such as liability and children’s use of chatbots.

But federal lawmakers have a hodgepodge of proposals. Mr Sanders and Mr Casar want a ban on any vaguely defined “superintelligent” AI. One bill in the House would require that AI firms use third-party evaluators for their models and allow the commerce department to suspend those posing imminent “catastrophic” risk; in an attempt to give structure to dread, the bill defines “catastrophic” as the death of more than 50 people or damages of at least $1bn.

A separate AI Safety Bill being negotiated in the Senate by John Thune, the Republican leader, Ted Cruz, a Republican senator, and Amy Klobuchar, a Democratic one, is thought to let the government seek a court order to ban models from being released in the first place. The toughest bill, from Josh Hawley, a Republican senator, and Richard Blumenthal, a Democrat, would force developers to submit AI models to the government for testing.

Even within the administration, there have been mixed messages. On September 15th, the day after Mr Trump dubbed attempts to curb AI a “SICK conspiracy”, J.D. Vance, the vice-president, said that AI firms “building Frankenstein…should stop” unilaterally. Firms would likely need an antitrust waiver to co-ordinate such a pause without fear of prosecution.

The Democrats’ debate contains its own contradictions. On September 14th Barack Obama, a former president, Kamala Harris, a former vice-president and presidential nominee, and Nancy Pelosi, a former House speaker, all endorsed Project Blueprint, an initiative to rebuild ties between the technology industry and Democrats, and to promote what Ms Harris called “sensible” oversight. However, it is bankrolled by Ron Conway, a venture capitalist who has also supported Leading the Future, the super PAC which helped water down state regulations.

On paper, many AI firms back some sort of regulation. Anthropic, the most enthusiastic, has cited the Federal Aviation Administration and Food and Drug Administration as models; in the meantime, it has said it would embed third-party evaluators in its lab, a practice that competitors claim they will follow. Google has suggested a body setting industry standards, modelled on the Financial Industry Regulatory Authority. But no lab has thrown its heft behind any of the big state or federal proposals in full, with the exception of Anthropic’s support for California’s law.

Voters’ patience for “sensible” regulation may wear thin. AI is already featuring heavily in the midterm elections this year. As the campaign for the presidency ramps up, after November, candidates’ positions may harden. “I predict to you every single Democratic candidate who runs in 2028 is going to be for a pause of some kind,” says Mr Carson. However, the next president will not take office until 2029. Any rules in the interim would need to be agreed by Congress and Mr Trump. While politicians debate, the models will evolve.■

资料来源｜英文原文《经济学人》（用户提供）