Quick answer
Economic Survey 2025-26 reports global surveyed-firm AI use rising from 50% in 2022 to 88% in 2025; India’s strategy stresses applications, skills, data and safety.
India’s AI strategy is an application-led development strategy, not simply a race to train the largest model. Revise the chapter through adoption evidence, structural trade-offs, shared infrastructure, human capital and accountable governance.
Sources: Chart XIV.1, paragraphs 14.2–14.6, 14.11–14.16 and 14.61–14.101; Box XIV.6. Adoption figures refer to global survey respondents, not all Indian firms.
Concept diagram
Recall the whole ecosystem, not just the model
An open revision notebook connects AI foundations and applications with governance, safety and sequencing.
- Compute is physical; adoption is economic; governance is institutional.
- Do not equate exposure with job loss, or open models with costless deployment.
Key numbers: keep populations and denominators attached
| Indicator | Value | Scope / qualification |
|---|---|---|
| AI in at least one business function, 2022 | 50% | Chart XIV.1; McKinsey respondent organisations |
| AI in at least one business function, 2023 | 55% | Chart XIV.1; not an Indian national estimate |
| AI in at least one business function, 2024 | 72% | Chart XIV.1; not full integration |
| AI in at least one business function, 2025 | 88% | Survey of 1,993 firms cited in paragraph 14.2 |
| Scaling AI across organisation | 31% | Among AI users in the cited 2025 survey |
| Fully deployed and integrated AI | 7% | Among AI users; not an economy-wide productivity statistic |
| Broadband access in India | Over 100 crore people | Wired or wireless; TRAI as of 31 December 2025, cited in paragraph 14.61 |
| Data-governance statutory foundation | DPDP Act, 2023 | Further categorisation and duties are Survey proposals |
Data chart
Headline adoption indicator: AI use in at least one business function
- Adoption rose by 38 percentage points between 2022 and 2025.
- Using AI in one function is much less demanding than full organisational integration.
Source: Economic Survey 2025-26, Chart XIV.1 (McKinsey, The State of AI in 2025). · These are reported global respondent shares, not Indian firm adoption rates. The 2025 survey cited 1,993 firms; do not assume an identical panel in all years.
Show the data (Per cent)
| Period / item | Respondent organisations using AI |
|---|---|
| 2022 | 50 |
| 2023 | 55 |
| 2024 | 72 |
| 2025 | 88 |
Twelve one-line facts for rapid recall
- Widespread AI use does not imply widely distributed frontier-model development capability.
- Frontier versus application development is a resource-allocation choice, not merely a prestige contest.
- Productivity gains and employment absorption can move differently in a labour-abundant economy.
- Open and interoperable systems support a bottom-up strategy; openness alone cannot certify safety.
- Compute infrastructure faces energy and water constraints as well as hardware-access limits.
- Sector-specific applications can align limited resources with concrete domestic problems.
- Human capital policy must accompany technology policy rather than follow displacement.
- The proposed AI Economic Council places labour realities and social stability within AI coordination.
- Box XIV.6 preserves cross-border openness while seeking enforceable accountability.
- A contemporaneous Indian mirror is not a requirement to process every dataset domestically.
- Safety evaluations should be ongoing, public and complemented by controlled red-teaming.
- The policy sequence is coordination first, capacity next and binding leverage last.
Terms: the minimum conceptual kit
Augmentation versus substitution
- Augmentation supports workers’ tasks; substitution replaces some human tasks with capital.
- Task exposure is not the same thing as a measured net employment loss.
Accountable portability
- Data can cross borders while remaining auditable, traceable and subject to intervention.
- Value retention adds domestic research, skills or productive capability benefits.
Red-teaming and sycophancy
- Red-teaming is controlled stress testing for misuse and failures.
- Sycophancy is excessive agreement that may boost engagement while worsening behaviour.
A Mains answer in four moves
Diagnose asymmetry
Separate broad adoption from concentrated frontier capability and bargaining power.
Choose domestic strengths
Prioritise application-led innovation, productive data use, human capital and shared infrastructure.
Govern proportionately
Combine graduated obligations, public access incentives and independent safety scrutiny.
Sequence for resilience
Coordinate experiments, scale with evidence and sustain hardware and skills adaptation.
Exam traps: distinguish the claim from the inference
Statements examiners like to twist — learn the correct version.
88% of Indian firms have fully integrated AI.
88% refers to surveyed organisations using AI in at least one function; full integration is a separate measure.
The rise from 50% to 88% is 38% growth.
It is 38 percentage points; percentage change uses the initial value as denominator.
All AI use is frontier-model development.
Application deployment and frontier training have different costs and capability requirements.
Data sovereignty means total physical containment.
The proposed framework favours accountable portability, mirrored oversight and domestic value retention.
A three-horizon roadmap establishes three statutory deadlines.
The chapter gives qualitative horizons, not fixed dates.
Prelims focus
- Revise Chart XIV.1 values with their respondent-based denominator.
- Pair Box XIV.6 with openness, oversight and domestic value retention.
- Distinguish model weights, training datasets and impact assessments.
Mains angle
- “India’s AI advantage lies in application-led innovation rather than scale for its own sake.” Discuss.
- Develop a balanced answer linking jobs, shared compute, domestic data, public safety evaluations and phased sectoral oversight.
Flip & recall
Answer in your head first, then tap the card to check.
Practice MCQs
Original Prelims-style questions with explanations.
0 / 3 correct · 3 left
Q1.Chart XIV.1’s 88% in 2025 refers to:
Q2.Moving from 50% to 88% represents an increase of:
Q3.Which pairing is correctly matched?
