Comprehension passage: Artificial intelligence, work and the social contract

Sample passage: Every previous wave of automation eventually created more jobs than it destroyed, but the comfort drawn from that record may be misleading when applied to artificial intelligence. Earlier machines amplified human muscle and freed workers to use their minds; the present generation of large models displaces cognitive tasks that were the bedrock of middle-class employment — drafting, translating, summarising, coding, advising. The transition is also unusually fast. Within four years of release, generative systems have moved from curiosity to daily tool in law firms, newsrooms, schools and clinics. Labour economists estimate that between a fifth and a quarter of measurable office tasks in advanced economies can already be automated; the share in routine clerical, paralegal and customer-service work is higher still. The Indian context is different but not insulated. The country's services boom rested on the export of cognitive labour at competitive prices, and any global compression of that demand will reach Indian cities first. At the same time, generative tools can lower the cost of producing tutors, paralegal aides, agronomic advice and primary-care triage, extending services to populations that the market has never reached. The social contract that emerged from the industrial age — schooling for the young, wages for adults, pensions for the old — was built on a predictable working life of roughly forty years. That contract is now under strain on three fronts: the half-life of vocational skills is collapsing, the boundary between employment and self-employment is dissolving, and the tax base that funds social protection is shifting from labour to capital. Adjustment will require a renewed compact: continuous learning entitlements, portable benefits, public investment in adoption support, and a fiscal architecture that taxes returns to data and algorithms. The question is no longer whether the machines will displace work, but whether democratic politics can update its instruments fast enough to keep the gains widely shared.

Model comprehension answers:

1. The optimism drawn from past technological transitions is misleading because earlier machines replaced muscle and freed workers for cognitive tasks, whereas generative artificial intelligence displaces precisely those cognitive tasks — drafting, summarising, coding, advising — that constituted middle-class employment. The pace of diffusion is also unusual: in four years the tools have moved from novelty to daily use in offices, schools and clinics, leaving little time for the gradual re-skilling that previous waves of automation historically allowed.

2. India is unusually exposed because its services-led growth has rested on exporting cognitive labour — back-office work, software development, business-process outsourcing — at competitive prices. Any global compression of demand for such labour will reach Indian cities first. At the same time, generative tools can extend tutoring, paralegal aid, agronomic advice and primary-care triage to populations the market has never reached, so the impact will be mixed and uneven rather than uniformly negative across the country.

3. The traditional social contract assumed schooling for the young, wages for adults and pensions for the old, calibrated to a working life of about forty years. Three pressures now weaken that compact: the half-life of vocational skills is shortening so rapidly that one period of schooling no longer suffices; the boundary between employment and self-employment is dissolving as gig work spreads; and the tax base is shifting from labour to capital, undermining the revenue that funds welfare. Each pressure compounds the others.

4. The author proposes a renewed compact with four elements. Continuous-learning entitlements would give every adult a publicly funded right to retrain across a long working life. Portable benefits, decoupled from a single employer, would protect workers whose careers cross many jobs and sectors. Public investment in adoption support would help small firms absorb the new tools without falling behind. A fiscal architecture taxing returns to data and algorithms would refill the revenue base that labour taxes alone can no longer sustain.

5. By 'whether democratic politics can update its instruments fast enough', the author asks whether the slow legislative and administrative cycles of democracies can match the speed at which artificial intelligence is reshaping work. Markets adjust quickly; political institutions, by design, do not. The sentence reframes the central question as one of governance velocity rather than of technological inevitability — the displacement is happening regardless, but the distribution of its costs depends on how rapidly policy can be revised to keep gains widely shared.