Quick answer
Economic Survey 2025-26 proposes an AI Economic Council with 5 principles, linking adoption to jobs, skills and readiness rather than speed alone.
The proposed AI Economic Council would coordinate AI deployment with India’s employment structure, education system and developmental priorities. Separate from the Governance Council discussed in Box XIV.5, it would ask whether the timing and manner of adoption improve human welfare, not simply whether the technology can be deployed.
Concept diagram
Give AI’s economic effects a common table
The proposed council is shown as a coordination table, not as an already-established institution.
- Economic spillovers cut across labour, infrastructure, industry and public services.
- Coordination should connect evidence, affected groups and policy feedback.
Basics: technological feasibility is not economic readiness
Automation versus augmentation
- Automation replaces tasks previously performed by people; augmentation helps people perform tasks better.
- A deployment can combine both effects, so jobs affected must be mapped rather than inferred from model capability alone.
Ex ante assessment
- An ex ante labour-impact assessment occurs before deployment and identifies likely displacement and transition needs.
- It differs from waiting for job losses and compensating afterwards.
Risk-weighted governance
- Oversight intensity should reflect likely harm and institutional readiness rather than treating every use identically.
- An incentive-based approach can encourage responsible adoption without assuming that voluntary action alone resolves all risks.
Box XIV.5: five proposed principles for the AI Economic Council
| No. | Principle | Operational implication |
|---|---|---|
| 1 | Human primacy and economic purpose | Require a credible pathway to net social and economic benefit, including jobs, productivity diffusion or public-service quality |
| 2 | Labour-market sensitivity by design | Assess informality, skill heterogeneity, regional variation and limited safety nets before deployment |
| 3 | Sequencing over speed | Match adoption to readiness across data, skills, legal frameworks and labour adjustment capacity |
| 4 | Co-evolution of technology and human capital | Pair technology policy with education, vocational adaptation, reskilling and credential recognition |
| 5 | Public-interest safeguards and ethical non-negotiables | Set boundaries around surveillance misuse, worker monitoring, discrimination and opaque decisions |
Three proposed readiness categories, not a blanket moratorium
Deploy now
Uses can proceed where readiness and safeguards support net benefits. The category is proposed in Box XIV.5, not a published universal whitelist.
Pilot
Use bounded experimentation where evidence or institutional readiness needs strengthening.
Defer
Delay uses when adjustment capacity, safeguards or supporting institutions are inadequate.
Survey comparison of governance approaches: context, not a current legal inventory
| Approach | Example in paragraph 14.53 | Lesson for India |
|---|---|---|
| Omnibus law | European Union Artificial Intelligence Act | A broad legislative framework is one option |
| Application-specific legislation | China | Rules may differ by application |
| Voluntary, non-binding guiding principles | United States, as characterised in the Survey | Not every governance approach begins with an omnibus statute |
| Indian institutional foundation | MeitY governance guidelines; proposed AI Governance Group and assisting technical committee | Build a light, incentive-based, risk-weighted approach adapted to national priorities |
What the Council would do in practice
A decade-long deployment roadmap
- Work with private firms to identify job profiles, geographies and the magnitude of automation and augmentation over the next decade.
- Connect the findings to education reform, transition policies and the pace of deployment rather than promising zero displacement.
Track deployment and learn
- The Survey calls for evolving transparency reporting and product registration requirements to help track adoption.
- Treat governance as continuous monitoring, learning and course correction, not a one-time technical approval.
Protect human intelligence
- In education, use AI to supplement teachers and learners rather than replace sustained reading, critical thinking and analytical writing.
- The Survey cites MIT- and Microsoft-linked studies on cognitive risks; this is not proof that every AI use causes permanent cognitive harm.
Exam traps: distinguish claim from inference
Statements examiners like to twist — learn the correct version.
The AI Economic Council is presented as an already operational statutory regulator.
Box XIV.5 is a proposal; it describes a coordinating role and distinguishes it from the Governance Council.
Sequencing over speed means refusing all AI adoption.
The proposed categories are deploy now, pilot and defer, depending on readiness.
Labour sensitivity requires ignoring productivity.
The proposal seeks productivity gains and net benefits without eroding inclusion, employment or dignity of work.
Prelims focus
- Box XIV.5 sets out 5 principles and 3 readiness categories.
- The proposed AI Economic Council is separate from the Governance Council.
- Labour-impact assessment is ex ante; the proposed deployment roadmap spans the next decade.
Mains angle
- Critically examine whether an AI Economic Council can reconcile productivity growth with labour absorption in a highly informal economy.
- Propose a governance cycle: assess impacts, sequence uses, build skills, track deployment and revise safeguards; distinguish coordination from duplicative regulation.
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.Which description best matches Box XIV.5?
Q2.Which set contains the proposed adoption categories?
Q3.Labour-market sensitivity by design requires:
