特朗普政府对人工智能和量子领域的关注对科学意味着什么

qimuai 发布于 阅读:15 一手编译

特朗普政府对人工智能和量子领域的关注对科学意味着什么

内容来源:https://www.sciencenews.org/article/trump-ai-quantum-science

内容总结:

据美国媒体报道,在特朗普政府第二任期内,美国科学界多个领域面临经费削减和项目延迟的困境,公共卫生和气候科学等研究受到冲击。然而,人工智能和量子科学等议题却获得了更为稳固的支持。

美国能源部负责科学事务的副部长达里奥·吉尔表示,人工智能、量子科学以及超级计算“将成为改变科学和工程研究方式的三大支柱”。2025年11月,特朗普签署行政令启动“创世纪任务”,这项耗资50亿美元的跨政府计划旨在通过人工智能推进科学研究,最初由能源部牵头,目前已扩展至15个以上联邦机构。同年6月,特朗普还签署行政令,旨在加速量子计算机、量子传感器和量子网络等工具的研发。7月,白宫科技政策办公室发布了一份123页的报告,描绘了联邦科学“黄金时代”的愿景。据白宫科技政策办公室下属机构统计,2025年联邦机构在人工智能研发上的支出约为42亿美元。

许多科学家对人工智能和量子研究带来的机遇充满热情,但也有人呼吁采取审慎态度。布朗大学物理学家、美国物理学会主席布拉德·马斯顿表示:“我最强烈的意见是,我希望看到所有科学领域都得到支持。 priorities必须有所选择,但保持良好平衡很重要,因为你永远不知道突破会来自哪里。”

通过“创世纪任务”,能源部计划利用其海量数据以及实验和计算设施,打造一套通用的人工智能工具和面向科学的共享生态系统。卡内基梅隆大学天文学家雷切尔·曼德尔鲍姆认为这一理念“在概念上非常强大”,若能按预期推进,有望推动科学技术取得重大进展。今年,能源部通过“创世纪任务”资助了近300个项目,其中包括曼德尔鲍姆团队领导的项目,旨在整合来自全球天文设施的庞大数据集,使人工智能模型能够基于这些数据进行训练,从而揭示隐藏的天文学洞见。其他项目还包括加速核反应堆开发、创建美国电网模拟以及研究核聚变等。美国国家科学基金会和国家卫生研究院等联邦机构也围绕该任务的国家科技挑战启动了各自的“创世纪”项目,涵盖健康、生物、能源、太空、国家安全等主题。

SLAC国家加速器实验室主任约翰·萨拉奥表示,许多科学家此前已在工作中使用人工智能,但“创世纪任务”确实创造了一个转折点。SLAC目前参与了20个“创世纪”项目,并牵头其中两个。

与人工智能相比,量子技术尚不成熟。量子经济发展联盟执行主任西莉亚·梅尔茨巴赫形容,在量子计算领域“我们仍处于马拉松的第一、二、三英里”。尽管如此,研究人员希望两种技术能够相互促进——人工智能可帮助设计和开发量子计算机,而量子计算在高效解决某些机器学习问题方面具有潜力。 accordingly,多个“创世纪”项目涉及量子纠错、量子传感器和量子材料。能源部的“量子创世纪”计划目标是到2028年开发出具有科学实用价值的纠错量子计算机。

人工智能和量子并非一直是热门领域。马斯顿指出,在深度神经网络发展之前,人工智能大多只是 curiosities,“这表明对科学资助采取平衡方法的重要性”。一个起初冷门的科学专业可能突然迸发出广泛的相关性。

在资源有限的背景下,优先发展某一研究领域必然意味着另一领域被降级。虽然“创世纪任务”中能源部部分的资金有一部分来自2025年通过的“一项大而美的法案”,但其余大部分来自能源部预算。曼德尔鲍姆虽然对“创世纪任务”感到兴奋,但也担心非人工智能相关工作的经费。她表示,其团队从能源部获得的一项宇宙学资助因不符合“创世纪任务”优先方向而在今年被削减。“如果现实是部分资金来自我所在领域的基础研究预算,那么我认为能源部就这些权衡与科学界沟通将是有益的。”

针对这些担忧,吉尔表示,能源部科学办公室向“创世纪任务”投入的资金迄今约4亿美元,与该办公室2026财年84亿美元的总预算相比规模较小。萨拉奥表示,在支持能源部核心研究的同时借助“创世纪任务”的势头很重要。他的印象是事情正朝着正确方向发展。“是否有真正令人兴奋的机遇?是的,百分之百。如果平衡把握不当是否存在风险?同样,百分之百。”

中文翻译:

在特朗普第二届政府任内,许多科学领域举步维艰,资金削减和项目延期困扰着公共卫生、气候科学等领域的研究。但有些领域却站稳了脚跟,包括人工智能和量子科学。

美国能源部负责科学事务的副部长达里奥·吉尔表示,这些技术连同超级计算,“将成为支撑未来科学与工程运作方式变革的三条腿”。

2025年11月,唐纳德·特朗普总统发布行政命令,启动“创世纪任务”——一项耗资50亿美元的全政府计划,旨在通过人工智能推进科学发展。该计划最初由能源部牵头,此后已扩展至超过15个联邦机构。

6月,总统发布行政命令,旨在加速量子计算机、量子传感器和量子网络等工具的研发。7月,白宫科技政策办公室发布了一份123页的报告,阐述了其对联邦科学“黄金时代”的愿景,即使各机构的研发投资与政府优先事项保持一致。

据白宫科技政策办公室负责量化此类投资的一个下属机构统计,2025年联邦机构在人工智能研发上的支出约为42亿美元。

许多科学家对人工智能和量子研究带来的机遇充满热情。但一些人提醒需要采取审慎的态度。“我最强烈的意见……是我希望看到所有科学领域都得到支持,”布朗大学物理学家、美国物理学会主席布拉德·马斯顿说。“优先级必须有所选择,但保持良好的平衡很重要,因为你永远不知道突破会来自哪里。”

通过“创世纪”计划,能源部打算利用其海量数据以及实验和计算设施,打造一套通用的人工智能工具和面向科学的共享生态系统。“这在概念上是一个非常强大的想法,”匹兹堡卡内基梅隆大学的天文学家蕾切尔·曼德尔鲍姆说。“如果它能按预期推进,可能会推动科学技术取得巨大进步。”

今年,能源部通过“创世纪”计划资助了近300个项目——其中包括曼德尔鲍姆领导的一个项目。她的团队致力于整合来自世界各地天文设施的庞大数据集,例如薇拉·鲁宾天文台、暗能量光谱仪和兹威基瞬变设施。这将使人工智能模型能够基于这些数据进行训练,从而有助于发现隐藏的天文学洞见。

其他项目包括加速核反应堆开发、创建美国电网模拟以及研究核聚变。包括国家科学基金会和国家卫生研究院在内的联邦机构也启动了各自的“创世纪”计划,围绕该任务的国家科学技术挑战展开,涵盖健康、生物学、能源、太空、国家安全等主题。

许多科学家早已在工作中使用人工智能。但“创世纪”计划“确实创造了一个拐点”,加州门洛帕克SLAC国家加速器实验室主任约翰·萨拉奥说。SLAC目前参与了20个“创世纪”项目,并牵头其中两个。

与人工智能相比,量子技术尚不成熟。在量子计算方面,“我们仍处于马拉松的第一、二或三英里处,”量子经济发展联盟执行主任西莉亚·梅兹巴赫说。该联盟是一个由公共和私人机构组成的联合体,致力于推动美国的量子技术发展。尽管如此,研究人员希望这两种技术能够相互促进。例如,人工智能可以帮助设计和开发量子计算机,而量子计算在高效解决某些机器学习问题方面展现出前景。

因此,若干“创世纪”项目涉及量子纠错、量子传感器和量子材料。能源部的“量子创世纪”计划旨在到2028年开发出具有科学实用价值的纠错量子计算机。

人工智能和量子并非一直是热门话题。“这两个领域……在取得近期进展之前都算是冷门,”马斯顿说。例如,在深度神经网络发展之前,人工智能基本上只是一种新奇事物。“这说明对科学资助采取平衡方式的重要性,”他说。起初只是一个冷门的科学专业,可能会突然爆发为广泛相关的领域。

在资源有限的背景下,优先发展一个研究领域必然意味着降低另一个领域的优先级。虽然“创世纪”计划中能源部承担的部分资金来自2025年通过的《一项大而美的法案》,但其余大部分来自能源部自身的预算。

尽管曼德尔鲍姆对“创世纪”计划充满热情,但她也担心非人工智能相关工作的资金问题。她说,她的团队从能源部获得的一项宇宙学资助与“创世纪”任务的优先方向不一致,今年被削减了。“如果现实情况是部分资金来自我所在领域的基础研究预算,那么我认为能源部就这些取舍与科学界进行沟通将是有益的,”曼德尔鲍姆说。

针对这些担忧,吉尔表示,从能源部科学办公室流向“创世纪”计划的资金——迄今约4亿美元——与该办公室2026财年84亿美元的总预算相比微不足道。

萨拉奥说,在保持“创世纪”计划势头的 Sicher 同时继续支持能源部的核心研究非常重要。他的印象是,事情正朝着正确的方向发展。“是否真的有令人兴奋的机遇?是的,百分之百,”萨拉奥说。“如果平衡把握不当,是否存在风险?同样,百分之百。”

英文来源:

Many fields of science have struggled under the second Trump administration, with funding cuts and delays plaguing research in fields such as public health and climate science. But some topics have found firmer footing, including artificial intelligence and quantum science.
These technologies, along with supercomputing, “are going to be the three legs of the stool that is going to transform how science and engineering is done,” says Darío Gil, undersecretary for science at the U.S. Department of Energy.
In November 2025, President Donald Trump issued an executive order to launch the Genesis Mission, a $5 billion whole-of-government effort to advance science via artificial intelligence. The initiative began with the DOE but has since expanded to more than 15 federal agencies.
In June, the president issued an executive order that aims to accelerate the development of tools such as quantum computers, quantum sensors and quantum networks. And in July, the White House Office of Science and Technology Policy issued a 123-page report describing its vision for a “golden age” of federal science, in which agencies align investments in research and development with administration priorities.
Federal agencies spent some $4.2 billion on AI R&D in 2025, according to an arm of the White House Office of Science and Technology Policy that quantifies such investments.
Many scientists are enthusiastic about the opportunities for AI and quantum research. But some caution that a measured approach is needed. “My strongest opinion … is that I want to see all fields of science supported,” says Brad Marston, a physicist at Brown University and president of the American Physical Society. “Priorities have to be chosen, but getting a good balance is important, because you never know where some advance will come from.”
Through Genesis, the DOE plans to use its troves of data as well as experimental and computing facilities to produce a common set of AI tools and a shared ecosystem for science. “That’s conceptually a very powerful idea,” says astronomer Rachel Mandelbaum of Carnegie Mellon University in Pittsburgh. “And if it pans out as intended, it could drive great advances in science and technology.”
This year, DOE funded nearly 300 projects through Genesis — including one led by Mandelbaum. Her team’s effort aims to unify immense datasets from astronomical facilities around the world, such as the Vera C. Rubin Observatory, the Dark Energy Spectroscopic Instrument and the Zwicky Transient Facility. That would enable AI models to be trained on the data, which could help uncover hidden astronomical insights.
Other projects include efforts to accelerate nuclear reactor development, create simulations of the U.S. electric grid and study nuclear fusion. And federal agencies including the National Science Foundation and the National Institutes of Health have launched their own Genesis efforts, centered around the mission’s National Science and Technology Challenges, which include topics in health, biology, energy, space, national security and more.
Many scientists were already wielding AI in their work. But Genesis “really did create an inflection point,” says John Sarrao, director of SLAC National Accelerator Laboratory in Menlo Park, Calif. SLAC is now involved with 20 Genesis projects and leads two.
Compared with AI, quantum technologies are less mature. With quantum computing, “we’re still at mile one or two or three of a marathon,” says Celia Merzbacher, executive director of the Quantum Economic Development Consortium, a coalition of public and private institutions that works to bolster quantum tech in the United States. Nevertheless, researchers hope the two technologies can amplify one another. AI could help design and develop quantum computers, for example, while quantum computing shows promise for efficiently tackling certain machine learning problems.
Accordingly, several Genesis projects involve quantum error correction, quantum sensors and quantum materials. And the DOE’s Quantum Genesis initiative aims to develop scientifically useful, error-corrected quantum computers by 2028.
AI and quantum were not always hot topics. “Both of those areas … were kind of backwaters until recent advances,” Marston says. Before the development of deep neural networks, for example, AI was mostly a curiosity. “It points to the importance of having a balanced approach to science funding,” he says. What starts as an obscure scientific specialty may suddenly explode into wide relevance.
In the context of limited resources, prioritizing one area of research necessitates deprioritizing another. While some funding for the DOE’s portion of Genesis came from the One Big Beautiful Bill passed in 2025, much of the rest came from the DOE’s budget.
As excited as Mandelbaum is about Genesis, she is concerned about funding for non-AI related work. A cosmology grant her group received from the DOE didn’t align with Genesis mission priorities and was reduced this year, she says. “If the reality is that some of this [funding] is coming out of the baseline research budget in my field, then I think that it would be beneficial for the DOE to engage the scientific community about those trade-offs,” Mandelbaum says.
In response to these concerns, Gil says that the money going to Genesis from the DOE’s Office of Science — about $400 million so far — is small compared to the office’s total FY2026 budget of $8.4 billion.
Sarrao says that it’s important to build on the momentum of Genesis while still supporting the DOE’s core research. His impression is that things are trending in the right direction. “Are there really exciting opportunities? Yes, 100 percent,” Sarrao says. “Are there risks if you get the balance wrong? Equally well, 100 percent.”

AI科学News

文章目录


    扫描二维码,在手机上阅读