AI 行业上周经历了糟糕的一夜。民主党在中期选举中大获全胜，这是对唐纳德·特朗普的谴责，他是该行业最有力的捍卫者。
The AI industry had a bad night last week. The Democrats’ sweeping midterms victory is a rebuke of Donald Trump, the industry’s most powerful champion.
几乎在所有可能的地方，选民都否决了数据中心：在九个州的 30 多个地方通过了暂停令和其他限制措施。而且，随着在得克萨斯州、俄亥俄州和佐治亚州等地获胜，估计有 41% 的规划数据中心产能位于当选州长支持暂停令的州。
Almost everywhere they could, voters rejected datacenters: moratoriums and other restrictions passed in more than 30 localities across nine states. And, with wins in places such as Texas, Ohio and Georgia, an estimated 41% of planned datacenter capacity is in states with governors-elect backing moratoriums.
诚然，选举结果主要并非关乎 AI。2024 年，特朗普以降低物价和不发动新战争为竞选纲领，随后却发动了一场推高物价的战争。根据 NBC 出口民调，虽然腐败、生活成本、伊朗和移民被视为对人们投票更为重要，但 23% 的选民将 AI 和数据中心列为其决定中的“关键”因素，民主党在这些选民中以 67% 对 30% 获胜。
To be sure, the election results were not primarily about AI. In 2024, Trump ran on lowering prices and starting no new wars, then began a war that raised prices. While corruption, the cost of living, Iran and immigration were seen as more important to people’s votes in NBC exit polls, AI and datacenters were cited by 23% of voters as “crucial” in their decision, and Democrats won these voters 67% to 30%.
AI 将成为 2028 年选举周期的决定性政治斗争。美国政治的未来将属于选民认为能够驯服该行业的一方。
AI will become the defining political fight of the 2028 cycle. The future of US politics will belong to whoever voters believe will bring the industry to heel.
按 AI 产业大亨自己的说法，他们正竞相制造可能毁灭我们的、摧毁就业的机器。Anthropic 首席执行官达里奥·阿莫代伊写道：“AI 不是特定人类工作的替代品，而是人类的通用劳动力替代品。”
AI industrialists, in their telling, are racing to build job-destroying machines that might also destroy us. Anthropic’s CEO, Dario Amodei, writes that “AI isn’t a substitute for specific human jobs but rather a general labor substitute for humans”.
而 OpenAI 首席执行官山姆·奥特曼曾臭名昭著地说：“AI 很可能、极有可能、某种程度上会导致世界末日，但与此同时会诞生伟大的公司。”正如一位微软高管所言，他们“利用人类历史上最大规模的劳动成果盗窃”构建了他们的机器。你能编出更好的反派吗？
And OpenAI’s CEO, Sam Altman, infamously said, “AI will probably, most likely, sort of lead to the end of the world, but in the meantime there will be great companies.”They built their machines on “the largest theft of labor in human history”, as one Microsoft executive put it. Could you script better villains?
近年来，没有哪个议题的关注度增长得比 AI 更快。全国各地的地方组织推动了反对 AI 数据中心的浪潮。在一个充斥着失控 AI 黑客事件和 AI 开发者发出显著警告的夏天之后，AI 灭绝人类的恐惧已进入主流视野。
And in recent years, no issue has grown more in salience than AI. Local organizing around the country drove an explosion of opposition to AI datacenters. Following a summer full of rogue AI hacking incidents and prominent warnings from AI developers, AI extinction fears have gone mainstream.
由人工智能驱动的黑客攻击和网络攻击近期激增，且将变得严重得多。而我们迄今看到的所有反弹，都发生在失业率为 4% 的时期。随着大规模劳动力自动化的迹象日益明显，我们会发现，反 AI 情绪的上限离我们还很远。
Hacks and cyber-attacks, driven by AI, have recently spiked and will get much worse. And all the backlash we’ve seen to date emerged in a period of 4% unemployment. As signs of widespread labor automation grow, we’ll find that we’re nowhere close to the ceiling of anti-AI sentiment.
要驯服这个行业很难，但这样做的政治收益巨大——而且谁都有机会。当 NBC 询问哪个党派在 AI 问题上会做得更好时，30% 的选民说是民主党，27% 说是共和党，43% 说两者都不是。较为精明的共和党人已意识到这一点，并正在为中期选举后与硅谷决裂铺垫理由。
The industry will be hard to tame, but the political upside of doing so is enormous – and it’s anyone’s game. When NBC asked which party would be better on AI, 30% of voters named Democrats and 27% named Republicans, with 43% saying neither. The savvier Republicans realize this and are already making the case for a post-midterm break with Silicon Valley.
确实，周二与试图用机器取代我们的行业关系过于亲密，成了共和党的政治包袱。特朗普遵循亿万富翁风险投资家大卫·萨克斯和马克·安德森的建议，对 AI 采取了“放任自流”的做法。
Indeed, coziness with the industry trying to replace us with machines was a political liability for Republicans on Tuesday. Trump, following advice from billionaire venture capitalists like David Sacks and Marc Andreessen, has taken a “let it rip” approach to AI.
科技高管的捐款让他们在特朗普就职典礼期间占据了国会圆形大厅的黄金座位，没有哪个行业为他的舞会捐款更多。今夏，总统在社交媒体发文称：“美国各地社区不想要数据中心的唯一理由，是如果它们想最终变得落后和贫穷。”而在该行业的游说下，几乎每一位众议院共和党议员都投票支持特朗普《一个宏伟美好的法案》中的一项条款，以联邦层面无任何替代法规为代价，推翻各州的 AI 监管——这是前所未有的举措。
Tech executive donations got them prime seating in the Capitol rotunda during Trump’s inauguration, and no industry has donated more to his ballroom. This summer, the president declared in a social media post: “The only reason that communities throughout the U.S.A. should not want Data Centers is if they want to end up being backwards and poor.”And at the industry’s behest, almost every House Republican voted for a provision in Trump’s One Big Beautiful Bill Act to override state AI regulations with no federal replacement – an unprecedented move. (Senate Republicans wisely voted to strip the measure.)
（参议院共和党人明智地投票删除了该条款。）所有这些以及更多内容，都成了民主党攻击广告的素材，超过 130 场联邦和州长竞选提及科技问责或 AI（2020 年至 2024 年间，仅有两条民主党阵营广告提及大型科技公司）。
All this and more was fodder for Democratic attack ads, with more than 130 federal and gubernatorial races mentioning tech accountability or AI (only two Democratic-aligned ads between 2020 and 2024 mentioned big tech).
要获得不止一个选举周期的提振，民主党必须真正兑现共和党未能提供的问责。特朗普不会出现在 2028 年的选票上，但 AI 行业没有任期限制。
To get more than a one-cycle bump, Democrats actually have to deliver on the accountability Republicans failed to provide. And Trump won’t be on the ticket in 2028, but the AI industry isn’t term-limited.
谁能最令人信服地对奥特曼、阿莫代伊、埃隆·马斯克、黄仁勋和马克·扎克伯格发起竞选挑战，谁在政治上获得的收益将远超因放弃超级政治行动委员会捐款而遭受的损失，而这些捐款的局限性周二也暴露无遗。
Whoever can most convincingly run against Altman, Amodei, Elon Musk, Jensen Huang and Mark Zuckerberg will gain far more politically than they’ll lose in foregone Super PAC contributions, the limits of which were also on display on Tuesday.
主要人工智能行业超级政治行动委员会大举投入的唯一一场有竞争力的竞选，以民主党人安妮·安德鲁斯在南卡罗来纳州获胜告终，她的竞选口号是“保护孩子，而非大型科技公司”。马斯克创纪录地花费近5亿美元却以失败收场：他在投入最多的五场参议院竞选中，共和党全军覆没。
The only competitive race the main AI industry Super Pac spent big in resulted in a loss to Democrat Annie Andrews in South Carolina, who ran on “protecting kids, not big tech”. Musk’s record-breaking spending of nearly half a billion dollars flopped: Republicans lost every one of the five Senate races he spent the most on.
所有这些都反对民主党回到奥巴马时代与硅谷的亲密关系。周二晚间清楚地表明，经济民粹主义是美国政治中最强大的力量——而这与对已变得过于强大的企业保持友好关系背道而驰。
All this argues against a return to Democrats’ Obama-era chumminess with Silicon Valley. Tuesday night clearly demonstrated that economic populism is the most potent force in American politics – and it’s incompatible with a friendly relationship with corporations that have already grown far too powerful.
值得庆幸的是，担忧人工智能危险的人士拥有比以往更多、权力也更大的捍卫者。民主党领导层和关键的新任委员会主席已发出信号，人工智能将成为他们的首要任务之一。而该党的左翼派别正走得更远。
Thankfully, those worried about AI’s dangers have more – and more empowered – champions than ever. Democratic leadership and key incoming committee chairs have signaled that AI will be one of their top priorities. And the party’s left flank is going further.
得克萨斯州民主党众议员格雷格·卡萨尔曾是一名劳工组织者，现任国会进步派核心小组主席，他宣称：“人工智能将成为2028年的头号议题。”卡萨尔与佛蒙特州参议员伯尼·桑德斯共同发起一项法案，该法案将暂停人工智能发展，直到联邦内阁级机构到位，并彻底禁止人工超级智能。
Texas Democrat Greg Casar, a former labor organizer and the chair of the Congressional Progressive caucus, declared: “AI will be the number one issue in 2028.”Casar has co-sponsored a bill with the Vermont senator Bernie Sanders that would pause AI development until a federal cabinet-level agency is in place and ban artificial super-intelligence outright.
在接受《60分钟》采访时，密歇根州当选参议员阿卜杜勒·埃尔-萨耶德呼吁对前沿人工智能开发实施“全面暂停”，并表示将把人工智能作为自己的头号议题。在胜选演说中，缅因州当选参议员特洛伊·杰克逊表示：“我们离《终结者》的场景已经不远了，我不想面临‘Hasta la vista, baby’（再见了，宝贝）的结局。” 拥有竞争性选情或白宫雄心的共和党参议员将希望与本党划清界限，这为民主党争取到突破60票阻挠议事门槛所需的选票创造了机会。
In his 60 Minutes interview, the Michigan senator-elect Abdul El-Sayed called for a “full-on pause” on frontier AI development and also said he was making AI his number one issue. In his victory speech, the senator-elect Troy Jackson of Maine said: “We are not that far from Terminator, and I don’t want to be ‘hasta la vista, baby.’”Republican senators with competitive races or White House ambitions will want to differentiate themselves from their party, creating opportunities for Democrats to pick up the votes needed to get over the 60-vote filibuster threshold.
而一些共和党人，如密苏里州的乔什·霍利，一直是起草强有力的人工智能监管法案的领军人物。特朗普一如既往地拥有否决权，但他释放出与民主党合作的惊人开放态度。
And some Republicans, like Missouri’s Josh Hawley, have been leaders on drafting robust AI-regulation bills. Trump, as he’s reminded us, could always use his veto power, but he’s signaled a surprising openness to working with Democrats.
为减轻人工智能带来的危害并兑现其降低生活成本的授权，民主党应当通过广受欢迎的亲劳工立法，迫使特朗普决定是否动用否决权。一些构想包括：禁止监视式定价和“老板软件”，让工人更容易组建工会（并跨行业组织），以及为人工智能开发者设立严格责任。
To mitigate the harms from AI and deliver on their affordability mandate, Democrats should pass popular pro-worker legislation and dare Trump to veto it. Some ideas: ban surveillance pricing and “bossware”, make it easier for workers to unionize (and organize across sectors), and create strict liability for AI developers.
民主党需要在改变企业激励结构方面发挥创造力，将导向从“替代劳动力”转向“增强劳动力”。立法应强化工人的权利，如果他们的工作成果被用于训练导致裁员或减少工时的人工智能模型，他们应能主张工资被盗。
Democrats need to get creative in shifting the incentive structures for companies from labor replacement to worker augmentation. Legislation should strengthen workers’ ability to make wage theft claims if their work is used to train AI models that lead to layoffs or reduction of hours.
如果你开发和使用的人工智能所做之事，若由人类做将构成犯罪——正如OpenAI和Anthropic的人工智能代理所表现出的那样——那么你应承担刑事责任。
And if the AI you develop and use does what would be criminal if a human did it, as AI agents from OpenAI and Anthropic have, you should be criminally liable.
我之前曾在此处及我的新书《淘汰》中论证，我们迫切需要停止构建通用替代劳动力机器的竞赛（业界称之为通用人工智能）。
I’ve previously argued here and in my new book, Obsolete, that we urgently need to stop the race to build universal labor-replacing machines (what the industry calls artificial general intelligence).
这需要比我们目前拥有的更强政治意愿。依我之见，如果公众真正了解业界意图为何，我们会立即叫停进一步开发。
That requires more political will than we currently have. In my view, if the public really understood what the industry was trying to do, we would shut down further development immediately.
国会听证会可以迫使在宣誓下分享重要的非公开信息。今年夏天，Casar就AI公司所负责的一系列黑客攻击事件向它们提出了尖锐问题，但得到的回答不完整且具有误导性。
Congressional hearings can compel the sharing of important non-public information under oath. This summer, Casar asked AI companies difficult questions about the spate of criminal hacking incidents they have been responsible for but got incomplete and misleading answers in return.
现在，民主党将能够用传票权力支持未来的调查——参议员Ed Markey曾向科技巨头高管提醒这一点：“他们将不得不为我们即将打印的所有传票种植新的森林。”国会听证会还可以围绕某些技术上公开但不为人所熟知的事情制造轰动效应，比如该行业的圣杯。AI公司明确计划用能够训练出更智能下一代的AI系统完全取代其研究人员，希望引发AI能力的爆炸，从而产生在几乎所有方面都超越人类的系统。
Now, Democrats will be able to back future inquiries with subpoena power – something the senator Ed Markey reminded big tech executives of: “They are going to need to plant new forests for all of the subpoenas we’ll be printing.”Congressional hearings can also create a spectacle around something that is technically public, but not widely known, like the industry’s holy grail. AI companies are explicitly planning to fully replace their researchers with AI systems that can train smarter generations, hoping to kick off an explosion in AI capabilities that results in systems that are superhuman at basically everything.
听起来有风险？别担心，他们还计划完全自动化AI安全研究。如果你与领导这项工作的人交谈，他们会承认这是个糟糕的计划——只是想不出更好的办法。
Seems risky? Don’t worry, they’re planning to fully automate AI safety research too. If you talk to the people leading this work, they’ll admit it’s a terrible plan – they just can’t think of a better one.
民主党应该把Altman、Amodei和Musk拉到世人面前，让他们解释他们到底在试图做什么，以及为什么在未经我们同意的情况下这样做。这也将是一个机会，让民主党展示他们会把谁的利益放在首位：是公众，还是试图用机器取代我们的亿万富翁。
Democrats should drag Altman, Amodei and Musk before the world and make them explain just what it is they’re trying to do and why they’re doing it without our consent. It will also be a chance for Democrats to show whose interests they’ll put first: the public or the billionaires trying to replace us with machines.
Garrison Lovely是一名自由记者，著有《过时：AI产业万亿级替代人类的竞赛——以及如何阻止它》。
Garrison Lovely is a freelance journalist and the author of Obsolete: The AI Industry’s Trillion-Dollar Race to Replace Us – And How to Stop It