雅思阅读 182: The Machines at the Office Desk（办公桌旁的机器）

改编从 World Bank / McKinsey / European Parliament（2025-2026年）。雅思阅读 Section 3 难度，约 1050 词。 素材来源：https://documents.worldbank.org/curated/en/099827011182513988/pdf/IDU-1300d27a-b3d3-43d9-8a52-047f784776c0.pdf

Reading Passage

A. For two centuries, the standard story of automation was told about the factory. First the loom, then the assembly line, then the industrial robot took over the routine physical tasks — and, in the process, created new jobs that required hands rather than routines. Office work was, until recently, assumed to be safe from this cycle, because reading, writing, judging and conversing were capacities that machines demonstrably lacked. The arrival of generative AI has overturned that assumption faster than almost anyone predicted. The people now most exposed to automation are not the welders and packers but the knowledge workers: marketers, accountants, paralegals, junior programmers, customer-service agents. A 2025 McKinsey survey found that 51 percent of organisations were already reducing their need for entry-level roles because of generative AI, and US Bureau of Labor Statistics data showed that unemployment among college graduates aged 23 to 27 had risen noticeably from its 2019 low. The first rung of the career ladder, in other words, is the rung that is disappearing first. This is a historically unusual pattern: previous waves of automation displaced routine physical labour while leaving entry-level cognitive jobs intact, so that a school-leaver could start as a clerk and learn the trade on the job. That apprenticeship path is now being eroded from below, just as tuition fees and housing costs have made the transition into work more expensive.

B. The evidence from job postings is sharper than opinion polls. A World Bank study, analysing millions of online vacancies across countries, estimated that by mid-2025 postings for occupations with above-median exposure to AI substitution had fallen by about 12 percent in the United States relative to postings in less-exposed occupations — and that the effect was intensifying over time, not fading. The decline was concentrated in high-income countries. In middle- and low-income countries, the same study found the impact was much smaller and, statistically, indistinguishable from zero. The reason is not that poorer countries are exempt; it is that their economies already employ fewer people in the writing, analysis and coding tasks that generative AI automates. The displacement risk, in other words, follows the offshored white-collar work — it lands first on the educated middle class of the rich world, not on the informal labourer in a village whose tasks have never been digitised in the first place.

C. Yet the picture is not simply one of mass unemployment. The World Economic Forum's 2025 Future of Jobs Report estimated that, by 2030, around 92 million jobs worldwide would be displaced by automation, but that a further 170 million new jobs would be created — a net addition of roughly 78 million. The new roles would not be identical to the old ones: demand for AI fluency, McKinsey found, jumped nearly sevenfold in the two years to mid-2025, and the ability to frame good questions and interpret AI outputs — rather than to perform routine analysis oneself — became a job requirement in occupations employing some seven million workers. At the same time, demand for purely "prompt-writing" skills is already being absorbed into the tools themselves, as the next generation of AI systems learns to clarify what the user actually meant. The European Parliament's briefing on AI and jobs concluded that over 70 percent of current skills can still be applied in both automatable and non-automatable work: most skills will not become obsolete, they will simply be used differently, with humans spending less time on common tasks and more time on judgement, care and relationships.

D. The uncomfortable part is the transition. It is one thing to say that skills will be "used differently"; it is another to retrain a 45-year-old paralegal into a prompt engineer, or to convince a graduate that the entry-level job they trained for no longer exists. McKinsey's more aggressive 2025 analysis suggested that currently demonstrated AI technologies could automate activities accounting for as much as 57 percent of US work hours, with software "agents" alone capable of performing tasks occupying 44 percent. Roles with the highest automation potential make up roughly 40 percent of total US employment, concentrated in legal and administrative services. That scale of change cannot be absorbed by individual reskilling alone. It requires a social architecture: unemployment systems that recognise that the displaced are not failures, education systems that teach adaptability rather than fixed craft, and employers willing to hire people on the basis of demonstrated potential rather than a prior CV in a role that no longer exists.

E. The long historical record offers both comfort and warning. When electricity was introduced into factories in the 1920s, it did not simply replace steam engines one-for-one; it allowed the whole layout of the plant to be redesigned, and productivity gains took twenty years to show up, because firms had to reorganise work around the new power source. Generative AI may follow the same pattern: the obvious task-replacement is visible today, but the larger productivity gains will come only when organisations redesign jobs, workflows and products around what the machines can actually do. Whether that redesign produces a better, more humane division of labour, or a pared-back workforce in which a small elite manages AI and everyone else scrambles for whatever human-only jobs remain, is not a technical question. It is a political one. The technology does not dictate the answer; it merely narrows the time in which societies must choose it. The machines, this time, are not on the factory floor. They are already, quietly, on the office desk. The choice facing societies is not whether to adopt the technology but whether the gains from it are broadly shared, and whether the workers whose tasks are absorbed are given the time, the retraining and the safety net to move on to what comes next.

Questions 1-4

Choose the correct heading for paragraphs B, C, D and E from the list of headings below.

List of Headings i. Why the new automation targets the office, not the factory ii. The uneven geography of job-posting decline iii. Displacement alongside new creation — the net employment picture iv. The transition problem and its social cost v. What the history of electricity teaches us vi. The ethics of prompt engineering vii. Why low-income countries are most affected first

1. Paragraph B: ____
2. Paragraph C: ____
3. Paragraph D: ____
4. Paragraph E: ____

Questions 5-8

Choose the correct letter, A, B, C or D.

5. Which workers are most exposed to generative AI, according to recent surveys? A. Factory welders and packers. B. Knowledge workers such as marketers, accountants and junior programmers. C. Bus drivers and dentists. D. Farm labourers.

6. What did the World Bank study find about job postings in the US by mid-2025? A. Exposed postings rose by 12 percent. B. Exposed postings fell by about 12 percent relative to less-exposed ones. C. There was no change in postings. D. All postings doubled.

7. According to the World Economic Forum, what is the net employment effect projected by 2030? A. A net loss of 92 million jobs. B. A net gain of about 78 million jobs. C. No change in total employment. D. A loss of 170 million jobs.

8. What lesson does the author draw from the history of electricity in factories? A. New technology replaces old machines instantly. B. Productivity gains arrive only after organisations redesign work around the new technology. C. Electricity caused mass unemployment in the 1920s. D. New power sources have no effect on factory layout.

Questions 9-13

Do the following statements agree with the claims of the writer?

Write:

• TRUE if the statement agrees with the information
• FALSE if the statement contradicts the information
• NOT GIVEN if there is no information on this

9. The World Bank found the negative job-posting effect was strongest in middle- and low-income countries.
10. Demand for AI fluency rose nearly sevenfold in the two years to mid-2025.
11. Currently demonstrated AI technologies could automate up to 57 percent of US work hours.
12. The author argues that generative AI will automatically produce a more humane division of labour.
13. Most new AI-related jobs are located in Silicon Valley.

Questions 14-15

Complete the summary below using NO MORE THAN TWO WORDS from the passage.

The new automation targets knowledge workers: McKinsey found that 51 percent of organisations were reducing entry-level roles because of generative AI. By mid-2025, US job postings in AI-exposed occupations had fallen by about (14) __________ percent. The long historical lesson from electricity is that productivity gains arrive only after organisations (15) __________ work around the new technology.

答案与解析

← 上一篇 | 返回雅思焦点 | 下一篇 →