The Financial Page

Can A.I. Be Pro-Worker?

人工智能能成为劳动者的盟友吗?
As fears of mass unemployment grow, three leading economists advocate some policies to shift the focus from job displacement to job enhancement.
随着对大规模失业的恐惧日益加剧,三位知名经济学家提出若干政策建议,试图将关注点从岗位替代转向岗位升级。
A human hand and a robotic hand work together on data analysis surrounded by graphs and charts.
Source photograph from Getty / NanoStockk

In recent weeks, remarkable things have been happening on Wall Street. As the major A.I. developers have been rolling out new versions of their models, and new work tools to sit atop them, investors have been knocking down the value of many big and profitable companies over fears that their businesses and employees will be disrupted, or displaced entirely. Hundreds of billions of dollars of value have been wiped out. Enterprise-software companies, like Salesforce and Workday; cybersecurity companies, like CrowdStrike; and wealth managers, such as Charles Schwab and Raymond James—they’ve all been hit. Early last week, selling extended to the broader market after Citrini Research, a little-known financial-research firm, posted a lengthy “thought exercise” about the impact of A.I., in which, by 2028, Citrini claims, soaring unemployment among white-collar workers will crimp consumer spending, and this will plunge the economy into a financial crisis and a recession.

最近几周,华尔街正发生着不同寻常的变化。随着各大人工智能开发商陆续推出新版模型及基于这些模型的新型工作工具,投资者们开始大幅抛售许多大型盈利企业的股票——他们担心这些公司的业务和员工将被颠覆,甚至被完全取代。短短数日,数千亿美元市值蒸发殆尽。从企业软件公司(如Salesforce和Workday)到网络安全公司(如CrowdStrike),再到财富管理机构(如嘉信理财和瑞杰金融),无一幸免。上周初,一家名不见经传的金融研究机构Citrini Research发布了一份关于人工智能影响的长篇“思想实验报告”,声称到2028年,白领失业率飙升将抑制消费支出,进而引发金融危机与经济衰退,这股抛售潮随即蔓延至更广泛的市场。

Later in the week, as other analysts poked holes in the Citrini scenario, the market recovered some of its losses. But the gyrations illustrate the power of two assumptions about A.I. that go largely unquestioned, on Wall Street and elsewhere: that the new technology is so powerful that it will transform the economy utterly; and that, despite being designed by humans, it’s now a force unto itself, whose progress can’t be reshaped or redirected. In short, we are all slaves to the A.I. algorithms and their inner workings, which remain somewhat mysterious even to their creators.

本周晚些时候,随着其他分析师对西特里尼(Citrini)的预测提出质疑,市场收复了部分失地。但市场的剧烈波动揭示出,关于人工智能的两大假设在华尔街及其他领域几乎无人质疑:这项新技术如此强大,将彻底重塑经济;尽管由人类设计,它如今已成为一种自主力量,其发展进程无法被重塑或转向。简而言之,我们都已成为人工智能算法及其内在运作的奴隶——即便对创造者而言,这些机制仍带着几分神秘。

When you think about it, this second assumption is both terrifying and ahistoric. In the paperback edition of their 2023 book, “Power and Progress: Our Thousand-Year Struggle Over Technology and Prosperity,” Daron Acemoglu and Simon Johnson, two M.I.T. economists, point out that the lesson of earlier economic transformations is: “you can’t stop technological change, but you can shape it.” In early British textile factories, women and children worked twelve-plus-hour days in unsanitary environments. It took the advent of factory legislation to shorten the workday and improve working conditions. And in many countries, including the United States, the rise of labor unions was a key factor in insuring that technology-driven productivity gains fed through to wage increases and expanded employment benefits as well as higher profits. The Treaty of Detroit, a five-year wage contract agreed upon by General Motors and the United Auto Workers in 1950, made this compact explicit.

细想之下,这第二个假设既令人恐惧,又违背历史。麻省理工学院经济学家达龙·阿西莫格鲁(Daron Acemoglu)和西蒙·约翰逊(Simon Johnson)在2023年著作《权力与进步:我们与技术繁荣的千年之争》平装版中指出,早期经济转型的教训是:“你无法阻止技术变革,但可以塑造它”。在早期的英国纺织厂里,妇女和儿童每天在肮脏的环境中工作十二小时以上。直到工厂立法出台,才缩短了工时并改善了工作条件。在许多国家,包括美国,工会的兴起是确保技术驱动的生产率提升不仅带来更高利润,同时转化为工资增长和就业福利扩大的关键因素。1950年通用汽车公司与美国汽车工人联合会达成的五年期工资协议《底特律条约》,正是将这一契约关系明确化的标志。

Acemoglu and Johnson are leaders in their field: they shared the 2024 Nobel Prize in Economics with a University of Chicago economist, James Robinson. In a new report for the Brookings Institution titled “Building pro-worker AI,” which Acemoglu and Johnson wrote with another noted M.I.T. economist, David Autor, they challenge the assumption of societal powerlessness in the face of A.I. They lay out a policy agenda designed to make sure that it acts as “a force magnifier for human expertise” rather than as a job killer. “We have a lot of agency, a lot of choice in shaping the future of technology,” Acemoglu told the MIT Sloan Management Review, “and different futures correspond to different winners and losers, different benefits, different costs, different productivities.”

阿西莫格鲁和约翰逊是各自领域的领军人物:他们与芝加哥大学经济学家詹姆斯·罗宾逊共同获得了2024年诺贝尔经济学奖。在为布鲁金斯学会撰写的新报告《构建亲劳动者的AI》中,阿西莫格鲁和约翰逊与麻省理工学院另一位著名经济学家戴维·奥托合作,挑战了面对人工智能时社会无能为力的假设。他们提出了一套政策议程,旨在确保人工智能成为“人类专业知识的赋能放大器”,而非就业杀手。“我们在塑造技术未来方面拥有很大的能动性、很多选择,”阿西莫格鲁在接受《麻省理工斯隆管理评论》采访时表示,“不同的未来对应着不同的赢家和输家、不同的收益、不同的成本、不同的生产力。”

As an example of how A.I. could be used in a pro-worker fashion, the report points to an Electrician’s Assistant (EA), developed by Schneider Electric, a French-based multinational company. When confronted by a tricky problem, the electrician feeds information and pictures into an assistant, which is a large language A.I. model. The assistant conducts a diagnosis and issues recommendations, in an iterative fashion, for how to fix the problem. It also helps the electrician file maintenance reports, and the paper cites evidence that the time spent on this task has been halved. “Tools akin to EA could be readily built to support many additional trade and modern craft workers, such as plumbers, building contractors, and health-care workers,” the report says.

报告以施耐德电气公司开发的“电工助手”为例,说明人工智能如何以支持劳动者的方式被运用。当电工遇到棘手问题时,可将现场信息和照片输入这个基于大语言模型的人工智能助手。助手会以迭代方式进行诊断,并提出维修建议。它还能协助电工填写维护报告——数据显示这项任务所需时间因此缩短了一半。报告指出:“类似'电工助手'的工具可以轻松开发,用以支持更多技术工种和现代工匠,比如水管工、建筑承包商和医护人员。”

This is an encouraging story, but how representative is it? For every example like the Electrician’s Assistant, there is one in which A.I. is already displacing jobs, or, at least, it’s being used as an excuse for big layoffs. Last week, Block, a financial-services platform, announced that it was getting rid of four thousand workers, out of ten thousand in total, on the ground that A.I. could do their jobs. Even in cases where companies have employed A.I. programs without engaging in mass layoffs, they have often been used to surveil and coerce workers rather than empower them. Amazon has an Associate Development and Performance Tracker program that it employs in its warehouses and always-on cameras that it deploys in its delivery vehicles, which are two notorious instances. Last week, Burger King said that it’s testing new A.I.-powered headsets, which can be used, among other things, to check whether its customer-service employees say “please” and “thank you.”

这故事令人鼓舞,但究竟有多少代表性?每出现一个像“电工助手”这样的案例,就会对应另一个案例——人工智能已经在取代工作岗位,或者至少被用作大规模裁员的借口。上周,金融服务平台Block宣布将裁减四千名员工(占员工总数的40%),理由是人工智能可以完成他们的工作。即便在那些尚未进行大规模裁员就部署人工智能系统的公司,这些技术也常被用于监控和胁迫员工,而非赋能于人。亚马逊仓库采用的“员工发展与绩效追踪系统”,以及配送车辆上永不关闭的监控摄像头,就是两个臭名昭著的例证。上周,汉堡王表示正在测试新型人工智能耳机,其功能包括检查客服人员是否说了“请”和“谢谢”。

The three M.I.T. economists don’t underestimate the scale of the challenge. “None of the big companies are pouring even a small fraction of their investment into developing A.I. as a pro-human, pro-worker tool,” Acemoglu said in his interview. To reorient things, he and his colleagues make a series of policy recommendations, including changing the tax laws, fostering competition in the A.I. sector, and giving workers a direct stake in A.I. One key proposal is for the government to use its financial power—both as a provider of research grants, and as a buyer and user of technology systems—to push the development of A.I. in a pro-worker direction. In the health and education sectors, for example, which together make up about twenty-five per cent of the nation’s G.D.P., government (at the federal and local level) is a major purchaser of tech products—a position it could use to demand the development of A.I. assistants that enhance workers’ capabilities. When I called up Autor last week, to ask him about the report, he cited the opportunity for A.I. assistants to help nurses carry out more demanding medical tasks, and to help teachers offer their students personalized support. “We pay for this stuff, we use it, the welfare of our children and grandchildren depends on it,” Autor said, referring to taxpayers. “I’m not saying the government should take over A.I., but it should use its power to shape its development.”

三位麻省理工学院的经济学家并未低估挑战的规模。“没有一家大公司投入哪怕一小部分资金,将人工智能开发成亲人类、亲劳动者的工具,”阿西莫格鲁在采访中表示。为了扭转局面,他和同事们提出了一系列政策建议,包括修改税法、促进人工智能领域的竞争,以及让劳动者直接参与人工智能的利益分配。其中一项关键提议是,政府应利用其财政力量——既作为研究资助方,也作为技术系统的购买者和使用者——推动人工智能朝着有利于劳动者的方向发展。例如,在医疗和教育领域(两者合计约占美国国内生产总值的25%),政府(联邦和地方层面)是科技产品的主要采购方,这一地位可用来要求开发能提升劳动者能力的人工智能助手。上周我致电奥托询问这份报告时,他提到人工智能助手有望帮助护士完成更具挑战性的医疗任务,并协助教师为学生提供个性化支持。“我们纳税人为此买单,我们使用这些技术,我们子孙后代的福祉都仰赖于此,”奥托说,“我并非主张政府接管人工智能,而是认为政府应运用其影响力引导其发展方向。”

In theory, the tax code could also be used to reshape the incentives of A.I. developers and users. When firms make investment decisions, they often have a choice of buying new labor-saving equipment, such as a chatbot, or hiring new workers and retraining existing ones. The current tax code, with its low rates on capital income and accelerated depreciation schedules, pushes businesses in the first direction. It “favors capital to an enormous extent, while it is very burdensome toward workers,” Autor pointed out. One way to change this would be to raise taxes on capital and reduce taxes on labor: that would make the code more neutral. A more drastic and more politically challenging option, which Autor said is worth considering, would be to tax consumption rather than working.

理论上,税收制度也可用于重塑人工智能开发者和使用者的激励结构。企业在投资决策时,常面临选择:是购置节省劳动力的新设备(例如聊天机器人),还是招聘新员工并对现有员工进行再培训。现行税法以较低的资本所得税率和加速折旧政策,将企业推向第一条道路。“这套制度对资本极度优待,却对劳动者施加沉重负担,”奥托尔指出。改变现状的一种方式是提高资本税、降低劳动税,使税制趋于中性。奥托尔认为更激进且政治阻力更大的方案也值得探讨——对消费而非劳动征税。

Part of the report that particularly caught my eye is a section titled “Discouraging expertise theft.” Right now, A.I. companies “freely scrape content from websites, social media, YouTube, newspapers, Wikipedia, and blogs, then statistically recombine this material and sell access to the results,” the report notes. “Authors, journalists, visual artists, musicians, translators, and countless other creators find their work appropriated as training data, with no compensation or control.” A recently published book, “The Means of Prediction,” by an Oxford economist, Maximilian Kasy, likens this grab to the enclosure of common land by landlords during medieval times—a development which greatly benefitted the landlords but destroyed the livelihoods of many small farmers. “A lot of the internet is being enclosed and resold to us as private property,” Autor said. “This is a huge reallocation of property rights.” With some firms using the performance of their own employees as data to train A.I. models, the report argues that the appropriation issue goes well beyond the internet: “Few employees would willingly train an apprentice designed to replace them, and yet this is precisely what happens when companies use worker expertise to build automation systems.”

报告中特别引起我注意的部分,是题为“遏制专业知识窃取”的章节。报告指出,当前人工智能公司“自由地从网站、社交媒体、YouTube、报纸、维基百科和博客抓取内容,然后对这些材料进行统计重组并出售访问权限”,“作家、记者、视觉艺术家、音乐家、翻译人员以及无数其他创作者发现自己的作品被挪用作训练数据,却得不到任何补偿或控制权”。牛津大学经济学家马克西米利安·卡西(Maximilian Kasy)在新近出版的《预测手段》一书中,将这种数据掠夺比作中世纪地主对公共土地的圈占——这种发展极大地惠及地主,却摧毁了许多小农的生计。“互联网的很大一部分正在被圈占,然后作为私有财产重新卖给我们,”奥托说,“这是一场巨大的产权重新分配。”报告指出,由于一些公司使用自家员工的绩效数据来训练人工智能模型,这种侵占行为已远远超出互联网范畴:“几乎没有员工会心甘情愿地培训一个旨在取代自己的学徒,然而当公司利用员工专业知识构建自动化系统时,这种情况恰恰正在发生。”

To address this issue, the report calls for new “legal frameworks that support workers’ ownership of their capabilities and creative output.” That sounds good, but what would it mean in practice? One model could be the book-publishing industry, where Anthropic has agreed to pay $1.5 billion to resolve a class-action lawsuit filed by authors and publishers for copyright infringement. (Like many writers, I have filed claims under this settlement.) But lawsuits are, at best, a very limited solution. The expertise of most cognitive workers—doctors, teachers, lawyers, consultants, accountants, software programmers, and so on—isn’t copyrighted. Neither are the physical skills that packers, drivers, builders, and other blue-collar workers possess, and which robotics companies are busy trying to replicate.

为解决这一问题,报告呼吁建立新的“法律框架,支持劳动者对其能力与创造性成果的所有权”。这听起来不错,但实践中意味着什么?图书出版业或许可提供一种参照模式——人工智能公司Anthropic已同意支付15亿美元,以和解作家与出版商提起的侵犯版权集体诉讼。(与许多作者一样,我已根据该和解协议提交索赔申请。)但诉讼至多只是非常有限的解决方案。大多数认知工作者——医生、教师、律师、顾问、会计师、软件程序员等——的专业知识并不受版权保护。包装工、司机、建筑工人等蓝领群体掌握的肢体技能同样不受版权保护,而这些正是机器人公司正竭力复制的领域。

This broader threat of appropriation and immiseration demands broader policy responses, and Autor suggested two of them: wage insurance for workers displaced by A.I. and a universal basic capital endowment. Under the first proposal, which Autor said is based on a trade-adjustment policy implemented by the Obama Administration, workers who are made redundant by A.I. and are forced to take lower-paying jobs, would receive a temporary federal wage subsidy. The proposed universal basic capital endowment would be made available to everyone. At birth, they would receive a government-funded investment account—the idea being that it could eventually grow large enough to provide them with a supplementary income source that wasn’t dependent on them supplying labor.

这种更广泛的攫取与贫困化威胁,需要更广泛的政策应对。奥托尔提出了两项建议:为被人工智能取代的工人提供工资保险,以及设立全民基本资本基金。第一项提议基于奥巴马政府实施的一项贸易调整政策,因人工智能而失业、被迫接受低薪工作的工人将获得联邦政府的临时工资补贴。而拟议的全民基本资本基金将面向所有人开放——每个新生儿都将获得一个政府注资的投资账户,其理念在于这笔资金最终可能增值到足以提供补充性收入,使人们不必完全依赖劳动谋生。

Proposals of this nature have long been popular on parts of the left, where they are sometimes referred to as asset-based redistribution. More recently, in the One Big Beautiful Bill Act of 2025, Donald Trump appropriated the appeal, if not the substance, of this approach by establishing tax-advantaged investment accounts for Americans born between January 1, 2025, and December 31, 2028, which the federal government will “seed” with a thousand-dollar contribution. To have a real impact on people’s lives and incomes, a universal basic capital endowment would have to be much larger than the Trump accounts, and it would likely have to be financed by raising taxes on large agglomerations of wealth. Autor didn’t get into financing details, but toward the end of our talk he pointed out two facts about the modern U.S. economy that were evident even before the onset of A.I.: wealth ownership is highly concentrated (the richest one per cent of households own more than thirty per cent of total wealth) and the share of national income accruing to workers rather than owners of capital is falling sharply (since 2000, it has dropped by about ten percentage points).

这类提案在左翼阵营的部分群体中一直颇受欢迎,有时被称为基于资产的再分配。最近,在《2025年宏大美好法案》中,唐纳德·特朗普借鉴了这种思路的吸引力(即便未采纳其核心实质),为2025年1月1日至2028年12月31日出生的美国人设立了税收优惠投资账户,联邦政府将向每个账户“注入”一千美元启动资金。若要对民众生活和收入产生实质影响,全民基本资本基金的规模必须比特朗普设想的账户庞大得多,且很可能需要通过提高对巨额财富的征税来筹措资金。奥托并未详述融资细节,但在我们谈话接近尾声时,他指出了现代美国经济中两个在人工智能兴起前就已昭然若揭的事实:财富所有权高度集中(最富有的1%家庭拥有超过30%的社会总财富),以及国民收入中劳动者而非资本所有者所占份额正在急剧下降(自2000年以来已下降约十个百分点)。

On its current course, A.I. seems likely to accentuate both these trends, with alarming implications for inequality, welfare, and democracy. It’s easy to question whether the proposals that Autor and his colleagues have put forward are adequate to address this challenge—he freely admitted that they don’t have an “omnibus solution”—but at least they are tackling the right questions. ♦

照此趋势发展,人工智能似乎很可能加剧这两种倾向,对不平等、社会福利和民主制度产生令人担忧的影响。人们很容易质疑奥托及其同事提出的方案是否足以应对这一挑战——他坦承他们并没有“一揽子解决方案”——但至少他们正在直面真正的问题。♦

2 free previews left — register to get 30 translations a month