共和党民调机构的一项新民调发现，爱荷华州和俄亥俄州的选民希望制定人工智能安全规则，即使美国并不总是在人工智能竞赛中处于领先地位。在美国中期选举前几天，该国最大的工会联合会表示正在起草自己的人工智能法案。
Voters in Iowa and Ohio want AI safety rules even if the US is not always first in the AI race, a new poll from a Republican pollster finds. Days before the US midterm elections, the country’s biggest union federation says it is writing its own AI bill.
由共和党民调专家罗布·施密特（Rob Schmidt）运营的Inquire公司，于10月19日至22日对爱荷华州501名可能投票者和俄亥俄州504名可能投票者进行了调查。Ashley Gold首先在Semafor上报道了结果。Inquire的民调备忘录给出的误差范围为各州4.4个百分点。两州都有参议院竞选，共和党正试图守住这些席位。
Inquire, a firm run by Republican pollster Rob Schmidt, surveyed 501 likely voters in Iowa and 504 in Ohio from 19 to 22 October. Ashley Gold reported the results first for Semafor. Inquire’s polling memo gives a margin of error of 4.4 points in each state. Both states hold Senate races that Republicans are trying to keep.
在两个州，67%的选民表示，美国应该通过合理的人工智能安全措施，即使它并不总是在全球人工智能竞赛中处于领先地位。在爱荷华州，24%的人倾向于将创新放在首位、减少规则，俄亥俄州为26%。
In both states, 67% of voters said the US should pass reasonable AI safety measures even if it is not always first in the global AI race. In Iowa, 24% preferred to put innovation first with fewer rules, and in Ohio 26% did.
十项规则，支持率均超70% 民调测试了十项提案，每项在两州获得的支持率均至少达到73%。防止最危险的人工智能落入敌手得分最高，爱荷华州为94%，俄亥俄州为93%。加强家长控制紧随其后，分别为90%和93%。
Ten rules, all above 70% The poll tested ten proposals, and each won at least 73% support in both states. Keeping the most dangerous AI out of enemy hands scored highest, at 94% in Iowa and 93% in Ohio. Stronger parental controls followed, at 90% and 93%.
部署前政府安全测试获得85%和87%支持。要求聊天机器人标明其为人工智能身份获得81%和85%支持。禁止聊天机器人与未成年人建立关系、禁止提供自残指示的支持率最低，为73%至77%。
Pre-deployment government safety testing drew 85% and 87%. Requiring chatbots to say they are AI drew 81% and 85%. Banning chatbot relationships with minors and banning self-harm instructions scored lowest, at 73% to 77%.
备忘录称，两党在人工智能本身的看法上存在分歧，但在规则问题上没有分歧。共和党对人工智能的看法比民主党更积极。尽管如此，共和党人选择安全措施而非减少监管，爱荷华州为50%对39%，俄亥俄州为56%对38%。
The two parties split on AI itself but not on the rules, the memo says. Republicans view AI more favourably than Democrats. Still, Republicans chose safety measures over fewer regulations, by 50% to 39% in Iowa and 56% to 38% in Ohio.
选民不确定信任哪个党。在爱荷华州，43%选择共和党，43%选择民主党。在俄亥俄州，共和党以43%对39%领先。
Voters are not sure which party to trust. In Iowa, 43% picked Republicans and 43% Democrats. In Ohio, Republicans led by 43% to 39%.
“倡导常识性安全措施的候选人将受益，而反对这些措施的候选人则面临政治反弹的风险，”Inquire在备忘录中写道。
“Candidates who champion common-sense safety measures stand to benefit, while those who oppose them risk political backlash,” Inquire wrote in the memo.
在两个州，超过十分之七的选民表示，如果参议院候选人支持 AI 安全措施，他们更有可能支持该候选人。不喜欢 AI 的选民多于喜欢的：爱荷华州为 37% 对 44%，俄亥俄州为 36% 对 44%。该备忘录未说明民调由谁出资。
More than seven in ten voters in both states said they would be more likely to back a Senate candidate who supports AI safety measures. Fewer voters liked AI than disliked it: 37% to 44% in Iowa and 36% to 44% in Ohio. The memo does not say who paid for the poll.
数据中心问题同样分化选民 关于数据中心的民调显示出类似倾向。Embold Research 本月为 Heatmap 进行的一项 Heatmap Pro 民调显示，71% 的受访者表示会反对在居住地附近建设数据中心，Robinson Meyer 为 Heatmap 撰文报道。
Data centres split voters Polling on data centres shows a similar mood. In a Heatmap Pro poll run by Embold Research this month, 71% of people surveyed said they would oppose a data centre near where they live, Robinson Meyer reported for Heatmap.
民调发现，共和党人的支持率略有回升。34% 的人会支持在附近建设数据中心，高于 8 月的 23%。
Support has slightly recovered among Republicans, the poll found. Some 34% would back a data centre nearby, up from 23% in August.
白宫的回应 白宫将重点放在保持美国领先地位上。唐纳德·特朗普总统将 AI 更名为“超级智能”，并称仍说“人工智能”的人是“敌人”。
The White House answer The White House has put the focus on keeping the US ahead. President Donald Trump has renamed AI “super intelligence” and called people who still say artificial intelligence “the enemy”.
“美国在超级智能领域的领导地位对我们的国家安全、经济繁荣和美国人民的生活质量至关重要，”白宫发言人 Liz Huston 告诉 Semafor。
“American leadership in super intelligence is essential for our national security, economic prosperity, and the quality of life of the American people,” White House spokesperson Liz Huston told Semafor.
“特朗普总统致力于确保美国继续在超级智能领域引领世界，同时保护美国消费者，”Huston 说。
“President Trump is committed to ensuring America continues to lead the world in super intelligence while protecting American consumers,” Huston said.
本月，其政府的 AI 工作组要求实验室报告其模型的失控行为。
This month, his administration’s AI task force told labs to report rogue behaviour by their models.
下届国会的工会法案 美国最大的劳工联合会 AFL-CIO 正在起草其称为“黄金标准”的 AI 立法。其主席 Liz Shuler 告诉 Semafor 的 Eleanor Mueller 和 Ashley Gold，希望议员们在下届国会通过该法案。
A union bill for next Congress The AFL-CIO, the largest US federation of labour unions, is drafting what it calls “gold-standard” AI legislation. It hopes lawmakers will pass it in the next Congress, its president, Liz Shuler, told Eleanor Mueller and Ashley Gold of Semafor.
Shuler 说，该法案将涵盖工人保护、隐私和数据保护，以及 AI 的经济影响。这将是该联合会的头号优先事项。
The bill will cover worker protections, privacy and data protection, and the economic impact of AI, Shuler said. It will be the federation’s top priority.
“职场正在发生的变化让人感到如此失控，[使]人们感到极度焦虑和不确定，”Shuler 说。
“The changes that are happening in the workplace that feel so out of control [have] people feeling extremely anxious and uncertain,” Shuler said.
她称公众对数据中心的抵制是焦虑情绪的表现形式。她预计，AI导致的失业将对政治产生更大的影响。
She called public pushback against data centres the way that anxiety is showing up. She expects job losses caused by AI to weigh on politics even more.
“数据中心将会建成。关键在于它们是否以优质的工会工作岗位和高标准的方式建设，”舒勒说。
“Data centers will be built. It’s a matter of if they’re built with good union jobs and a high-standard approach,” Shuler said.
该联合会的AI原则已获得两党支持者。4月，共和党众议员布莱恩·菲茨帕特里克和民主党参议员马克·凯利牵头致信支持这些原则。信中要求政府将工人保护纳入联邦AI政策。
The federation’s AI principles already have backers in both parties. In April, Republican congressman Brian Fitzpatrick and Democratic senator Mark Kelly led a letter backing them. It asked the administration to build worker protections into federal AI policy.
在俄亥俄州，共和党参议员伯尼·莫雷诺指责Anthropic的达里奥·阿莫代伊发表“危言耸听”的AI言论。中期选举将于11月3日举行。
In Ohio, Republican senator Bernie Moreno has accused Anthropic’s Dario Amodei of ‘alarmist’ AI talk. The midterms are on 3 November.