根据美国广播公司（ABC News）、《华盛顿邮报》与益普索（Ipsos）联合发布的一项最新民意调查，多数美国成年人对人工智能的发展持负面看法，同时反对继续大规模建设数据中心。调查显示，57%的受访者认为人工智能对社会弊大于利，66%反对在美国建设新的数据中心；如果数据中心建在自己所在的社区，反对比例进一步升至72%。
这项调查于2026年10月9日至13日进行，通过益普索的KnowledgePanel概率抽样调查平台访问了2545名美国成年人，整体误差范围约为正负2个百分点。调查结果显示，人工智能技术发展速度不断加快，但公众对其潜在收益与风险的看法正在出现明显分化。
在对人工智能整体社会影响的评价中，57%的受访者认为AI对社会有害，42%认为它有益。认为AI对社会“极其有害”的比例达到17%，而认为其“极其有益”的仅有2%。这表明，公众的负面评价不仅体现在一般性的担忧上，也有相当一部分人对人工智能的发展持强烈怀疑态度。
不同年龄群体对AI的看法存在差异。30岁以下的受访者中，65%认为人工智能对社会有害；30至39岁群体中，这一比例为59%；40岁及以上人群中，持相同看法的比例为53%。年轻人对AI的悲观程度更高，可能与他们对就业前景、职业发展和个人生活受到技术影响的担忧有关。
政治立场同样影响受访者的态度。民主党支持者中，63%认为AI对社会有害；无党派人士中，这一比例为58%。共和党支持者的看法则相对接近，51%认为AI有益于社会，47%认为其有害。
值得注意的是，当问题从社会整体影响转向个人体验时，调查结果出现了微妙变化。50%的受访者认为AI对自己有益，49%认为它对自己有害。也就是说，部分美国人虽然担心AI给社会带来的长期影响，却仍然认可它在个人生活中的实用价值。不过，认为AI对自己“极其有害”的受访者占15%，是认为其“极其有益”者比例的五倍。
调查还详细询问了公众对人工智能潜在风险的判断。73%的受访者认为，AI发展可能导致安全计算机系统遭到攻击，构成较高风险。这是调查中受到关注程度最高的风险之一，反映出公众对AI被用于网络攻击、入侵关键系统以及实施复杂网络犯罪的担忧。
就业问题同样引发广泛关注。65%的受访者认为，人工智能有较高风险导致大量工作岗位消失。调查显示，这种担忧横跨不同年龄、政治立场和收入群体，并非只集中在某一类劳动者身上。
此外，62%的受访者认为，人们可能逐渐失去独立作出决定的能力，越来越依赖人工智能提供的建议。54%担心AI会加剧经济不平等，认为技术带来的收益可能更多集中于少数企业和高收入群体，而普通劳动者未必能够平等分享生产力提升所创造的价值。
对于更极端的风险，33%的受访者认为人工智能的发展可能带来人类灭绝的高风险。虽然这一比例低于网络安全、就业和个人自主性等风险的担忧程度，但仍意味着约三分之一的美国成年人将AI视为可能威胁人类长期生存的技术。
与风险认知相比，公众对AI能够带来的好处预期相对有限。34%的受访者认为，AI很有可能让日常事务变得更加容易；30%认为AI有较大机会帮助美国在人工智能技术领域取得主导地位。只有20%认为AI很有可能提高人们的生产力，另有38%认为AI几乎没有或完全没有提高生产力的可能。
对于就业创造，公众的预期更加谨慎。仅13%的受访者认为AI有较大机会创造大量新工作岗位，而52%认为这种情况几乎不可能发生。调查结果反映出一种明显的担忧：公众更容易看到AI可能取代现有工作的风险，却尚未普遍相信新技术能够创造足够多的新岗位来弥补损失。
在监管问题上，约三分之二的美国人认为，联邦政府目前没有对人工智能的发展和应用实施足够监管。民主党支持者中，这一比例达到80%；无党派人士为66%；共和党支持者也有54%持相同看法。只有22%的受访者认为，政府当前的监管力度大致适当。
美国总统特朗普对AI的处理方式同样受到较多批评。调查显示，66%的受访者不认可特朗普处理人工智能问题的方式，只有30%表示认可。民主党支持者和无党派人士的不认可比例分别达到92%和75%。在共和党内部，特朗普的支持者与其他共和党人之间也存在差异，MAGA阵营支持者的认可比例明显更高。
与此同时，公众对两大政党能否妥善处理AI问题普遍缺乏信心。26%的受访者更信任民主党处理人工智能问题，19%更信任共和党，但有49%表示两党都不值得信任。与2026年4月的调查相比，公众对两党处理AI问题的信任度出现了变化，但不信任任何一方的比例仍然接近一半。
数据中心建设是此次调查的另一项重点。随着AI模型训练和推理需要大量计算资源，美国科技企业正在加速建设大型数据中心。然而，这些设施需要消耗大量电力和冷却用水，并可能影响当地电网负荷、居民能源账单以及周边社区环境，因此引发越来越多争议。
调查显示，66%的美国成年人反对在全国范围内建设新的数据中心。若建设地点位于受访者自己的社区，反对比例则升至72%。支持在全国范围内建设更多数据中心的受访者仅占30%。
这种反对态度并不局限于某个政治阵营。对于在自己所在社区建设新数据中心，民主党支持者的反对比例为81%，无党派人士为76%，共和党支持者也达到58%。即使是在支持人工智能产业发展的共和党群体中，仍有多数人不愿意在居住地附近建设新的数据中心。
不同地区的反对程度也有所差异。农村地区居民反对在本地建设新数据中心的比例达到78%，郊区居民为72%，城市居民为69%。这表明，即使数据中心能够为当地带来投资和税收收入，居民仍然可能担心其对电力供应、生活环境和社区质量造成影响。
数据中心问题也逐渐成为美国政治争论的一部分。特朗普政府强调，建设更多数据中心有助于推动美国经济增长，并在全球人工智能竞争中保持优势。但部分居民、地方官员和政策制定者则担心，数据中心扩张带来的收益与成本并不一定由同一群体承担。企业可以获得算力资源和商业收益，而当地居民却可能承担电网扩容、能源成本上涨以及环境压力。
此次调查呈现出的核心矛盾是：人工智能产业正在获得越来越多的资本和基础设施投入，但美国公众并未普遍接受这种扩张方式。许多人承认AI可能让部分日常任务更方便，却同时担心就业流失、网络安全风险、经济不平等和个人自主性下降。数据中心建设争议则将这些抽象担忧转化为更加具体的地方问题。
对科技企业而言，未来的挑战可能不再只是争夺芯片、建设算力设施和开发更强大的模型，还包括如何证明AI能够为普通人带来实际收益，以及如何降低其对就业、能源供应和社区生活的负面影响。对于美国政府而言，调查也显示，公众对人工智能监管的需求正在扩大，如何在产业竞争、技术创新和风险控制之间取得平衡，可能成为未来政策讨论的重要议题。
本文由 AI图鉴自动抓取并翻译，版权归原作者所有；如需查看原文请点下方链接。
According to a recent public opinion survey by the American Broadcasting Corporation (ABC News), the Washington Post and the Ipsos joint publish, most adults in the United States have negative views on the development of artificial intelligence, while opposing the continued large-scale construction of data centres. The survey showed that 57 per cent of respondents considered artificial intelligence to be more beneficial to society, and 66 per cent opposed the construction of a new data centre in the United States; if the data centre was built in their own community, the opposition rate would rise further to 72 per cent.
The survey, which was conducted from 9 to 13 October 2026, visited 2545 adults in the United States through the Knowledge Panel Probabilities Sampling Platform in Whipso, with an overall error range of about two percentage points positive or negative. The results of the survey show that the pace of development of artificially intelligent technologies is accelerating, but that public perceptions of their potential benefits and risks are becoming evident.
In the evaluation of the overall social impact of artificial intelligence, 57 per cent of respondents considered AI to be socially harmful and 42 per cent considered it useful. The proportion of AI considered to be “extremely harmful” to society reached 17 per cent, while only 2 per cent considered it “extremely beneficial”. This suggests that the negative public perception is not only reflected in general concerns, but also in the strong sceptical attitude of a significant proportion of the population towards the development of artificial intelligence.
There are differences among age groups regarding AI. Sixty-five per cent of those under 30 considered artificial intelligence to be socially harmful; 59 per cent of those aged 30 to 39; and 53 per cent of those aged 40 and over. Young people are more pessimistic about AI, which may be linked to their concerns about employment prospects, career development and the technological impact of personal life.
Political positions also influence the attitudes of interviewees. Sixty-three per cent of Democrats consider AI to be socially harmful; 58 per cent of non-partisans. The views of Republican supporters were relatively close, 51 per cent considered AI to be beneficial to society and 47 per cent considered it harmful.
It is noteworthy that when the problem shifts from the overall impact of society to the experience of individuals, the results of the survey have changed subtlely. Fifty per cent of respondents considered AI to be beneficial to themselves and 49 per cent considered it harmful to themselves. That is to say, some Americans, while concerned about the long-term impact of AI on society, still recognize its practical value in personal life. However, 15 per cent of respondents who considered AI to be “extremely harmful” to themselves were five times as likely to consider it “extremely useful”.
The survey also asked the public in detail about the potential risks of artificial intelligence. Seventy-three per cent of respondents felt that AI development could lead to attacks on secure computer systems, which posed a high risk. This is one of the highest-profile risks in investigations, reflecting public concerns about AI being used for cyberattacks, intrusion into critical systems and the commission of complex cybercrime.
The issue of employment has also attracted widespread attention. Sixty-five per cent of respondents felt that the high risk of artificial intelligence led to the disappearance of large numbers of jobs. The survey shows that this concern cuts across age groups, political positions and income groups and is not focused solely on one category of workers.
In addition, 62 per cent of respondents felt that people were at risk of losing their ability to make independent decisions and were increasingly relying on the advice provided by artificial intelligence. Fifty-four per cent feared that AI would increase economic inequality, suggesting that the benefits of technology might be more concentrated in a few enterprises and high-income groups, and that the general workforce might not be able to share equally in the value created by productivity gains.
For the more extreme risks, 33 per cent of respondents believed that the development of artificial intelligence could pose a high risk of human extinction. While this percentage is lower than the concern about risks such as cybersecurity, employment and personal autonomy, it still means that about one third of adults in the United States view AI as a technology that could threaten the long-term survival of humankind.
Public expectations of the benefits of AI are relatively limited compared to risk perceptions. Thirty-four per cent of respondents felt that AI was likely to make day-to-day business easier; 30 per cent felt that AI had a greater chance of helping the United States gain dominance in the area of artificial intelligence technology. Only 20 per cent believe that AI is likely to increase productivity, while another 38 per cent believe that AI has little or no potential to increase productivity.
Public expectations for job creation are more cautious. Only 13 per cent of respondents felt that AI had greater opportunities to create a large number of new jobs, while 52 per cent felt that this was almost impossible. The results of the survey reflected a clear concern that while the public was more aware of the risk that AI might replace existing jobs, there was still no widespread belief that new technologies could create enough new jobs to compensate for losses.
On the issue of regulation, about two thirds of Americans believe that the federal government currently does not have adequate regulation of the development and application of artificial intelligence. The proportion of Democrats is 80 per cent; non-partisans 66 per cent; and Republican supporters 54 per cent share the same view. Only 22 per cent of respondents felt that the current level of government regulation was largely adequate.
The treatment of AI by United States President Trump has also received much criticism. The survey showed that 66 per cent of respondents did not accept the way Trump dealt with artificial intelligence, and only 30 per cent did. The percentage of non-acknowledged Democrats and non-partisans reached 92 per cent and 75 per cent, respectively. Within the Republican party, there are also differences between Trump's supporters and other Republicans, with a significantly higher percentage of Maga's supporters.
At the same time, there was a general lack of public confidence in the ability of the two parties to deal properly with AI issues. Twenty-six per cent of the respondents had greater confidence in the Democratic Party to deal with artificial intelligence issues and 19 per cent in the Republican Party, but 49 per cent indicated that neither party was trustworthy. Compared with the April 2026 survey, there has been a change in the level of public confidence in the handling of AI issues between the two parties, but the rate of mistrust of either side remains close to half.
Data centre construction was another focus of the survey. As AI model training and reasoning require significant computing resources, United States science and technology enterprises are accelerating the construction of large data centres. However, the need for these facilities to consume large quantities of electricity and water for cooling and could affect local power grid loads, household energy bills and the surrounding community environment has given rise to increasing controversy.
The survey showed that 66 per cent of United States adults were opposed to the construction of new data centres throughout the country. If the site of construction is located in the respondent ' s own community, the percentage rises to 72 per cent. Only 30 per cent of respondents supported the construction of additional data centres throughout the country.
This opposition is not limited to a certain political camp. For the construction of new data centres in their own communities, 81 per cent of Democrats, 76 per cent of non-partisans and 58 per cent of Republican supporters were opposed. Even within the Republican community, which supports the development of artificially intelligent industries, the majority are still reluctant to build new data centres near their places of residence.
The extent of objections varies from region to region. The proportion of people in rural areas opposed to the construction of new local data centres amounted to 78 per cent, 72 per cent in the suburbs and 69 per cent in the cities. This suggests that even if data centres can generate local investment and tax revenues, residents may still be concerned about their impact on the supply of electricity, the living environment and the quality of communities.
Data centre issues are also becoming part of United States political debate. The Trump Government emphasized that building more data centres would help boost United States economic growth and maintain an advantage in the global competition for artificial intelligence. However, some residents, local officials and policymakers are concerned that the benefits and costs of data centre expansion are not necessarily borne by the same group. Enterprises have access to computing resources and commercial gains, while local residents may bear the burden of grid expansion, rising energy costs and environmental pressures.
The central contradiction that emerged from the survey was that the artificial intelligence industry was receiving increasing capital and infrastructure inputs, but that expansion was not widely accepted in the United States. Many acknowledge that AI may make part of its day-to-day tasks easier, while at the same time fearing job losses, cyber-security risks, economic inequality and a decline in individual autonomy. The data centre construction dispute translates these abstract concerns into more specific local issues.
For science and technology enterprises, the challenges ahead may no longer be just to compete for chips, build computing facilities and develop stronger models, including how to demonstrate that AI can deliver tangible benefits to ordinary people and how to reduce its negative impact on employment, energy supply and community life. For the United States Government, surveys have also shown that public demand for artificial intelligence regulation is growing, and how to balance industrial competition, technological innovation and risk control may be important topics for future policy discussions.
Auto-fetched and translated by AI Atlas. All rights belong to the original author.