ARTICLE · 1015909
经济学人|AI真的影响就业吗?「电脑看更佳」
高三版｜上海高考英语外刊精读
The jobs apocalypse is postponed. An AI jobs boom is here
就业末日推迟了：AI就业繁荣已经到来
一、读前导入
话题背景： 关于人工智能和就业的讨论常集中在裁员新闻，但判断一项技术对劳动市场的整体影响，还需要同时观察它创造了多少新岗位、基础设施投资带来了什么需求，以及生产率提高后是否会扩大服务需求。阅读时要特别区分局部受损行业与整个经济的净就业变化。 上海卷影子考点： 重点训练从多组就业数据中判断净效应，识别estimate、appears、may等限定表达，区分“某些岗位减少”和“整体就业崩溃”，并理解基础设施、AI专门岗位和生产率三条就业创造渠道。 | 读前问题： 1. 如果新闻中不断出现企业因AI裁员的消息，这是否足以证明AI正在减少整个经济体的就业？还需要哪些数据？ 2. 一种技术提高劳动生产率后，为什么有时反而可能增加而不是减少某些职业的就业？ |
二、外刊文章
P1When artificial intelligence began spreading through offices, many forecasts focused on mass unemployment. Yet the American labour market has not behaved that way so far. In August the economy added 162,000 jobs, while unemployment stood at 4.1%, lower than in almost 90% of months over the past half-century. Even younger workers, often expected to be among the first victims of AI, have not suffered a clear collapse. The gap between unemployment among 20-24-year-olds and the overall rate is close to a multi-decade low. P2That does not mean the technology is harmless. Hiring in professional and business services is running about 10% below its 2015-19 average. Large technology firms are trimming headcounts① as they reorganise around AI, while some smaller companies are replacing people with bots. American employers have announced about 16,000 AI-related job cuts a month on average this year. Those losses are painful, but they sit inside a churning labour market in which employers shed② roughly 1.7m workers in a typical month. P3The other side of the ledger is job creation. Spending on data centres and electricity has increased demand for construction and infrastructure workers; AI startups are hiring; established firms are creating new AI roles; and higher productivity may raise demand for some professional services. Adding these channels together, The Economist estimates that AI has so far created around 1m jobs in America. That figure appears③ to exceed the roughly 200,000 lay-offs attributed to AI since mid-2023 and may offset④ weaker hiring in many back-office roles. P4Infrastructure is a major part of the boom. Goldman Sachs calculates that annual spending on chips, servers, data centres, cooling and power is roughly $500bn higher than in 2022. Data-centre construction alone is running above $75bn a year. Employment in five industries closely tied to this build-out has risen by about 320,000 more than broader construction and manufacturing trends would suggest since 2023. Employers are also paying a premium⑤: some installation and maintenance jobs at data centres advertise wages around 40% above comparable work elsewhere. P5AI is also creating white-collar occupations. Engineers build models, data annotators judge outputs, “forward-deployed” engineers adapt systems for customers and newly created heads of AI decide how companies should use the technology. One estimate suggests that roughly 1% of professional jobs are now AI-specific⑥. LinkedIn identifies about 640,000 such new jobs between 2023 and 2025. The Economist also finds that engineers, software developers, mathematicians and data scientists have added roughly 730,000 jobs above the broader professional trend, though not all can be attributed⑦ to AI. P6A third channel is productivity. If AI lets lawyers draft contracts faster or analysts search filings in minutes, lower costs could increase demand enough to support more work. Some occupations once thought especially vulnerable⑧ have expanded: between 2023 and 2025 employment among paralegals rose about 11% and among market-research analysts 6%, against a national average near 2%. Yet disruption is real elsewhere. Customer-service employment has fallen about 10% since January 2023 and employment among secretaries and administrative assistants by roughly 15%, reflecting exposure to routine⑨ tasks. P7So far, the evidence points to AI as a net⑩ job creator rather than a jobs apocalypse. New forms of work are also likely to keep emerging as the technology improves. More than 6m Americans already work in computer-related occupations that did not exist before computers, and another 8m work in newer industries such as e-commerce, the gig economy and content creation. Tomorrow’s workers may supervise autonomous⑪ agents or settle disputes between them. Such jobs may sound fanciful⑫ today, but so did being an influencer not very long ago. | 重点词汇 / Key Words 01headcounts /ˈhedkaʊnts/n. 员工总数；人员规模 ──────────── 02shed /ʃed/v. 削减；失去 ──────────── 03appears /əˈpɪəz/v. 似乎；看来 ──────────── 04offset /ˈɒfset/v. 抵消；弥补 ──────────── 05premium /ˈpriːmiəm/n. 额外溢价；高于通常水平的部分 ──────────── 06AI-specific /ˌeɪˈaɪ spəˈsɪfɪk/adj. 专门与人工智能相关的 ──────────── 07attributed /əˈtrɪbjuːtɪd/v. 归因于；认为由……造成 ──────────── 08vulnerable /ˈvʌlnərəbl/adj. 容易受到影响或伤害的 ──────────── 09routine /ruːˈtiːn/adj. 常规的；重复性的 ──────────── 10net /net/adj. 净的；扣除相反影响后的 ──────────── 11autonomous /ɔːˈtɒnəməs/adj. 自主运行的 ──────────── 12fanciful /ˈfænsɪfəl/adj. 看似异想天开的；不切实际的 ──────────── |
三、文章结构理解
核心主线：失业预言对照现实 → 局部裁员确认 → 三条创造就业渠道 → 基建就业展开 → 白领新岗位展开 → 生产率与职业分化 → 净创造与未来新岗位 | |
1. 预言与现实 / Forecast versus reality 第1段用整体劳动力数据与此前的预测对照，功能是设置反差，不是给出结论。 | 2. 真实冲击 / Genuine disruption 第2段主动承认反面证据并给出数量级参照，作用是限定全文主张的适用范围。 |
3. 净效应转折 / The other side 第3—4段转入第一条解释路径，用投资额、岗位数与工资变化三类数据支撑同一条因果链。 | 4. 新职业形成 / New occupations 第5段转向另一类劳动力市场，说明一类新出现的岗位，与前一段构成并列而非递进。 |
5. 生产率渠道 / Productivity channel 第6段提出第三条机制，并在段末用两个方向相反的职业对照，提示该机制的作用并不均匀。 | 6. 历史视角 / Longer-term outlook 第7段先给出阶段性净判断，再把时间尺度拉长到技术史，以举例而非论证方式收尾。 |
作者态度 / Author attitude：判断依据：分别数出作者用于描述冲击的动词和用于描述新增就业的动词，再看第7段用的是 so far、appears、may 还是确定语气；净判断出现在哪一句要自己定位。/ Judge by weighing the disruption verbs against the job-creation verbs, then checking the tense and modality of the net claim. |
｜阅读抓手：不要只统计“裁掉多少岗位”，而要建立三栏：AI造成的岗位损失、AI直接或间接创造的岗位、尚属推测的未来效应；遇到estimates、appears、may、could时保留原文的不确定程度。
四、重点词汇
No. | Word | 中文 / 语境 | 同义替换 / 搭配 | 写作迁移表达 |
1 | net job creation | 净就业创造，指新增岗位扣除消失岗位后的总体效果，适用于技术与劳动市场议题 | overall employment gain; positive employment balance | evaluate net job creation |
2 | labour-market disruption | 劳动市场受到技术或经济变化冲击，适用于自动化和产业转型 | employment disruption; workforce upheaval | manage labour-market disruption |
3 | infrastructure investment | 基础设施投资，适用于经济增长、能源和数字技术话题 | capital spending on infrastructure; infrastructure expenditure | stimulate jobs through infrastructure investment |
4 | above trend | 高于原有趋势水平，适用于经济和数据分析 | beyond the expected trend; higher than baseline growth | grow above trend |
5 | productivity gain | 生产率提升，适用于技术、企业和宏观经济话题 | efficiency improvement; productivity increase | turn productivity gains into wider demand |
6 | routine task | 重复性、标准化任务，适用于自动化和职业变化 | repetitive task; standardised work | automate routine tasks |
7 | occupational shift | 职业结构变化，适用于劳动力转型 | change in job composition; workforce shift | prepare workers for an occupational shift |
8 | create new demand | 创造新的需求，适用于生产率、消费和就业关系 | generate demand; expand the market | create new demand for services |
9 | job displacement | 岗位被技术或产业变化替代，适用于AI与自动化议题 | worker displacement; employment loss | reduce the costs of job displacement |
10 | emerging occupation | 新兴职业，适用于未来工作和职业规划 | newly developing role; new profession | train for emerging occupations |
11 | weigh the evidence | 综合权衡不同证据，适用于数据型议论文和研究评价 | assess the evidence; consider all the data | weigh the evidence before predicting job losses |
12 | adapt to technological change | 适应技术变化，适用于教育、职业发展和产业转型 | adjust to new technology; respond to technological shifts | help workers adapt to technological change |
13 | protect workers during transition | 在产业转型期间保护劳动者，适用于就业政策和社会保障 | support displaced workers; cushion transition costs | protect workers during transition |
14 | avoid technological determinism | 避免认为技术会自动决定社会结果，适用于科技与社会议论文 | reject inevitable-outcome thinking; avoid deterministic claims | avoid technological determinism |
五、文章翻译
P1当人工智能开始在办公室中普及时，许多预测都集中在大规模失业上。然而，到目前为止，美国劳动力市场并没有这样发展。8月份，美国经济新增16.2万个岗位，失业率为4.1%，低于过去半个世纪近90%的月份。即使是常被认为会最先受到AI冲击的年轻劳动者，也没有出现明显崩溃。20至24岁人群失业率与总体失业率之间的差距接近数十年来的最低水平。
P2这并不意味着这项技术没有伤害。专业和商业服务行业的招聘量比2015—2019年的平均水平低约10%。大型科技公司正在围绕AI重组业务并削减人员规模，一些较小企业则用机器人替代员工。今年到目前为止，美国企业平均每月宣布约1.6万个与AI有关的裁员。这些损失对相关劳动者而言很痛苦，但它们发生在一个高度流动的劳动力市场中：典型月份里，雇主大约会削减170万个岗位。
P3账本的另一面是就业创造。数据中心和电力方面的支出增加了建筑和基础设施工人的需求；AI初创企业正在招聘；现有企业正在设立新的AI岗位；而生产率提升也可能增加某些专业服务的需求。把这些渠道加在一起，《经济学人》估计，AI迄今已在美国创造约100万个岗位。这一数字似乎超过了自2023年年中以来约20万个被归因于AI的裁员，而且可能抵消许多后台岗位招聘疲软的影响。
P4基础设施是这轮就业繁荣的重要组成部分。高盛估算，芯片、服务器、数据中心、冷却和电力方面的年度支出比2022年高约5000亿美元。仅数据中心建设每年的规模就超过750亿美元。自2023年以来，与这一建设热潮密切相关的五个行业，就业增长比更广泛的建筑业和制造业趋势所暗示的水平高出约32万个。雇主还在支付工资溢价：一些数据中心安装和维护岗位的招聘工资比其他地方类似工作高约40%。
P5AI也在创造新的白领职业。工程师构建模型，数据标注人员评价输出结果，“前线部署”工程师为客户调整系统，而新出现的AI负责人则决定公司应如何使用这项技术。一项估计认为，如今大约1%的专业岗位属于专门与AI相关的工作。LinkedIn统计出2023年至2025年间大约新增64万个此类岗位。《经济学人》还发现，工程师、软件开发人员、数学家和数据科学家的岗位增长比专业职业整体趋势高出约73万个，不过这些增长并不能全部归因于AI。
P6第三条渠道是生产率。如果AI让律师更快起草合同，或者让分析师几分钟内搜索完财务文件，那么更低的成本可能使需求增加到足以支持更多就业。一些曾被认为特别容易受到AI影响的职业反而有所扩张：2023年至2025年，律师助理就业增长约11%，市场研究分析师增长6%，而全国平均增幅接近2%。不过，其他地方的冲击是真实存在的。自2023年1月以来，客户服务岗位减少约10%，秘书和行政助理岗位减少约15%，这反映出这些职业较多涉及AI越来越擅长的重复性任务。
P7到目前为止，证据显示AI更像是一个净就业创造者，而不是带来所谓的“就业末日”。文章还预计，随着技术进步，新的工作形式将继续出现。目前已有超过600万美国人从事计算机出现之前并不存在的计算机相关职业，另有800万人在电子商务、零工经济和内容创作等较新的行业工作。未来的劳动者或许会管理自主智能体团队，或者处理这些智能体之间的争议。这样的职业今天听起来可能异想天开，但不久以前，“网红”同样如此。
｜使用建议：翻译只作理解辅助，精读时优先回到英文原文定位关键词和句子结构。