Agentic AI and the Future of Work, Exam-Ready Notes

Two documents separated by three weeks converged on the same warning. On 7 January 2025, the World Economic Forum’s Future of Jobs Report calculated that by 2030 artificial intelligence and structural change will create one hundred and seventy million jobs and displace ninety-two million — a net gain of seventy-eight million that hides enormous churn. Then, on 31 January, the Economic Survey 2024-25 devoted an entire chapter, Labour in the AI Era, to India’s exposure, calling for robust institutions to manage what it called an AI arms race already under way. Between those two documents sits the newest object of the debate: agentic AI — systems that do not merely answer but act, pursuing goals with autonomy over tools and workflows.

This card, dated twelve days after the Survey’s tabling, assembles the exam kit: what distinguishes agentic from generative AI, the global jobs arithmetic and its fine print, the Survey’s specific warnings for the Indian workforce, the IT-services exposure that makes this an Indian rather than abstract question, and the skilling-institution response taking shape. It is the workforce update to the technology cards this series has already built — AI governance belongs to the earlier cards; this one is about who works, and how, when software begins to act.

From Generative to Agentic

The definitional line exams will draw.

1. Generative AI. Systems that produce content — text, code, images — in response to prompts; the ChatGPT-era technology whose defining trait is that it answers.
2. Agentic AI. Systems that pursue goals: given an objective, they plan steps, call tools, browse, execute transactions and self-correct — the defining trait is that they act.
3. The autonomy gradient. Agents are graded by how much human approval each step requires — from suggest-only agents to fully autonomous loops — the taxonomy regulators and auditors increasingly borrow.
4. Why 2024-25 is the inflection. Tool-calling, memory and reasoning models matured together, letting start-ups and enterprise vendors ship computer-use and workflow agents at scale — the arrival that moved the labour debate from content generation to task execution.
5. The exam line. Generates versus acts; the autonomy gradient; 2024-25 as the deployment inflection — the three-clause definition that separates prelims bands.

The Global Arithmetic

What the Future of Jobs Report actually said.

1. The headline. Of roughly one billion jobs surveyed-equivalent, one hundred and seventy million new roles are expected by 2030 against ninety-two million displaced — the net positive seventy-eight million that newspaper headlines carried.
2. The churn beneath. Nearly one in five workers in the surveyed base sees their job transformed or gone — churn, not net loss, is the real story for policy.
3. The skills clock. Thirty-nine per cent of core skills change by 2030, and sixty-three per cent of employers already identify the skills gap as the biggest barrier to transformation — the two percentages that anchor every mains answer on skilling.
4. The growth table. AI and big data top the fastest-growing skills list, followed by networks, cybersecurity and technological literacy — while manual dexterity, reading and clerical roles head the declining column.
5. The exam line. 170 up, 92 down, net 78; 39 per cent skills change; 63 per cent gap — the four-number spine.

The Economic Survey’s Warning

What India’s own document added.

1. The chapter. Labour in the AI Era, in the Economic Survey 2024-25 tabled on 31 January 2025 — the first time the pre-Budget document devoted a full chapter to technology’s labour shock.
2. The anxiety finding. The Survey recorded that anxieties about AI’s effects run high in India, and that an arms race dynamic — its phrase for the deployment speed race among frontier labs, naming OpenAI explicitly — compresses the adaptation window societies have.
3. The vulnerability map. India’s workforce concentrates in exactly the zones agents reach first — routine cognitive work, back-office processing, content and customer interaction — including the entertainment industry’s creative trades the Survey called out for early exposure.
4. The institutional ask. Robust institutions — education, skilling and social-safety structures — rather than fragmented programmes, as the Survey’s answer to disruption it judged neither imminent catastrophe nor benign neglect.
5. The exam line. First-ever AI-labour chapter, arms-race caution, services-and-creative vulnerability, institutional demand — the four-point Survey summary.

The Indian Exposure

Why this is a services-economy question.

1. The IT-services weight. India’s flagship export industry — software services and business-process outsourcing — sells precisely the routine-cognitive labour that agentic systems automate first, making the Indian stakes structurally different from commodity economies.
2. The demographic clock. The same decade AI churns skills is the decade India’s demographic dividend must be encashed — the collision that elevates a technology story into a development one.
3. The services share. More than half of gross value added comes from services; even manufacturing employment is thick with clerical, logistics and coordination tasks agents can absorb.
4. The geography of work. Tier-two and tier-three cities hosting global capability centres mean disruption and opportunity will both be widely distributed — not confined to Bengaluru and Hyderabad corridors.
5. The exam line. IT-services concentration, demographic dividend overlap, services-heavy GVA, GCC spread — the four-fibre Indian weave.

The Response Architecture

Institutions and missions already on the board.

1. IndiaAI Mission. Approved in March 2024, its pillars include capacity-building alongside compute and datasets — the flagship through which AI skilling funds now flow.
2. The credit framework. The National Credit Framework and the National Skills Qualifications Framework let training be credited and stacked formally — plumbing the Survey’s robust-institution ask requires.
3. PMKVY 4.0. The fourth phase of the skills mission explicitly includes new-age and digital-skill courses — the delivery vehicle at district level.
4. Curriculum motion. School and university syllabi adding AI literacy and the University Grants Commission’s guidelines on AI in education — the supply-side response moving faster than most realise.
5. The exam line. IndiaAI capacity pillar, NCrF crediting, PMKVY 4.0, curriculum reforms — the response stack to name in any twenty-marker.

Governance and Work Adjacent

The regulatory frame this card inherits.

1. The advisories lineage. Deepfake-labelling advisories of 2024 under the IT Rules and the AI-governance consultations they seeded are the domestic regulatory thread agents will pull next — accountability when an agent, not a person, completes the transaction.
2. Global pointers. The European Union’s AI Act entered into force in 2024 with risk-tiered obligations; the Bletchley-to-Seoul summits built the safety-dialogue track — the two external references Indian answers should deploy.
3. The accountability question. When an autonomous system errs — a payment executed, a message sent — liability, audit and disclosure rules decide whether trust scales; governance is thus a labour question too, because adoption follows trust.
4. The federal angle. Education and labour are concurrent and state subjects respectively — implementation of any skilling wave runs through state capacity, a GS2 connection examiners reward.
5. The exam line. Advisory lineage, EU risk tiers, agent liability, concurrent-list delivery — the governance four.

How Exams Ask This Card

Question shapes and their marking engines.

1. Definition split. Agentic versus generative AI — the two-word answer that separates prepared candidates in prelims.
2. Data deployment. Quoting 170-92-78 with the thirty-nine and sixty-three percentages — the numbers that turn an essay into evidence.
3. India-specific evaluation. Assess the impact of agentic AI on India’s services-led employment — the twenty-marker this card drafts, exposure, Survey caution, response architecture as the paragraphs.
4. Institution design. Design the robust institutions the Survey demands — the answer that earns top band by specifying credit stacking, portable social security and tripartite oversight rather than repeating skilling.
5. Integration question. Connect the demographic dividend to the AI-skills clock — the cross-paper link between this card and the population-and-development syllabus.

Quick Revision: Ten Lines

One glance before the hall.

1. Definition. Agentic AI pursues goals with autonomy over tools; generative AI only produces content.
2. WEF 2025. By 2030: 170 million jobs created, 92 million displaced — net positive 78 million beneath massive churn.
3. Skills clock. Thirty-nine per cent of core skills transform by 2030; 63 per cent of employers flag the skills gap.
4. The Survey. Economic Survey 2024-25, tabled 31 January 2025, carried the chapter Labour in the AI Era.
5. The warning. The Survey cautioned against an AI arms race compressing adaptation windows, naming the front-runners’ pace.
6. The exposure. India’s IT-services, business-process and creative work sits in the first automation zone of agents.
7. The entitlement. The demographic dividend must be encashed in the very decade AI reshuffles skills — the central collision.
8. The missions. IndiaAI Mission capacity pillar, PMKVY 4.0 and the National Credit Framework form the response stack.
9. The governance. Liability, audit and disclosure for autonomous action remain open files riding on the 2024 advisory lineage.
10. The exam angle. Churn not collapse, institutions not slogans, exposure not abstraction — the three framing sentences.

Conclusion: Institutions Ahead of the Wave

The honest reading of early 2025 is that the numbers point to churn, not collapse — but churn at the speed of an arms race, in the exact labour markets where India earns its keep. The World Economic Forum’s net-positive arithmetic and the Economic Survey’s institutional demand are two halves of one instruction: growth from agentic AI is likely, capture of that growth is not guaranteed, and the difference will be made by education, credit and social-security institutions built before, not after, the wave lands. That is why this card ends where the Survey began — with robust institutions as the heroine of the answer. For a country whose demographic advantage and whose services engine share the same decade, the agentic-AI question is finally a simple one: will the institutions mature faster than the technology. The prepared candidate holds both. It is the workforce chapter of the technology ledger this series has been building — the governance and safety chapters live in the earlier AI cards.