1. India’s Productivity Paradox in the Age of AI
India’s official productivity indicators present a picture of stability. Output per worker rises slowly, wages grow incrementally, and labour productivity appears broadly unchanged. On the surface, this suggests continuity in how work is organised and performed.
However, this statistical calm contrasts with changes visible inside workplaces. Across offices, IT services, call centres, and government departments, work is being completed faster, coordination cycles are shorter, and errors are detected earlier. These micro-level changes reshape daily work processes.
The disconnect highlights a growing productivity paradox. Traditional metrics do not fully capture changes in the quality, speed, and reliability of work. If this gap is ignored, policy assessments risk underestimating structural transformation underway in the economy.
When productivity measurement fails to reflect changes in how work is done, governance decisions risk being based on incomplete signals.
2. Limits of Conventional Productivity Measurement
Most productivity metrics were designed for an industrial economy where output increased through more labour, more machines, or longer working hours. Intelligence was scarce and directly tied to human effort.
Artificial intelligence does not conform to this model. It provides cognitive support—drafting, summarising, debugging, and decision assistance—without adding labour hours or physical capital in the traditional sense.
As a result, AI-driven efficiency gains often pass through organisations without registering clearly in GDP or labour productivity data. Ignoring this limitation can lead to premature conclusions about the effectiveness of digital technologies.
Measurement systems shape perception; outdated metrics can obscure real but unconventional productivity gains.
3. High AI Adoption, Modest Aggregate Productivity Gains
The mismatch between adoption and measured productivity is not unique to India. In the United States, labour productivity growth has remained around 1–1.5 per cent annually for over a decade, with no visible surge even after widespread adoption of generative AI post-2022.
By 2024, nearly one-third of US workers were already using generative AI tools at work. Despite this rapid diffusion, aggregate productivity indicators show limited acceleration.
This tension fuels scepticism about AI’s economic impact. However, the issue may lie less with AI’s effectiveness and more with how its benefits are aggregated and recorded.
High adoption without headline productivity gains signals a measurement lag, not necessarily a technology failure.
Statistics:
- US labour productivity growth: ~1–1.5 per cent
- US generative AI adoption (2024): ~33 per cent of workers
4. Task-level Productivity Gains and Organisational Use
At the level of individual tasks, AI’s impact is significant. Field experiments in customer support show average productivity gains of around 14 per cent, rising to over 30 per cent for newer or less-experienced workers. Response quality improves, error rates fall, and employee attrition declines.
Studies of management consultants similarly report faster task completion and smoother coordination. These gains, however, are often absorbed internally rather than converted into higher measurable output.
Firms use AI to shorten turnaround times, reduce training costs, standardise outputs, and stabilise workflows. Consequently, output volumes remain constant even as processes improve.
AI raises efficiency within firms before expanding output; early gains are internalised rather than monetised.
Impacts:
- Task productivity gains: ~14 per cent
- Gains for newer workers: >30 per cent
- Reduced errors and attrition
5. India’s Experience: Faster Work, Stable Output
India’s IT services sector illustrates this pattern clearly. Large firms report 20–30 per cent reductions in coding and testing time in selected workflows through generative AI adoption.
Despite these efficiency gains, billed effort and overall revenue growth remain subdued. Faster delivery is absorbed into fixed-price contracts, margin protection, and quality improvements rather than expanded volumes.
From the perspective of national accounts, little changes. Productivity gains appear muted, even as internal work processes are significantly altered.
When efficiency gains are absorbed rather than scaled, measured productivity remains flat despite real improvements.
Statistics:
- Reduction in coding/testing time: 20–30 per cent
6. Time-saving, Not Labour-saving: A Key Blind Spot
AI primarily saves time rather than labour. It accelerates drafting, searching, summarising, and decision-making, reducing uncertainty and coordination costs.
Productivity statistics, however, are built to capture output per hour worked, not improvements in decision quality or speed. Consequently, gains that manifest as smoother workflows remain invisible in official data.
Another blind spot lies in averages. AI’s strongest effect is often on reducing variation—raising the performance of lower-skilled workers, reducing errors, and making outcomes more predictable. These distributional improvements matter for firms but escape aggregate metrics.
What productivity measures ignore—reliability and consistency—often matters most for organisational performance.
7. Historical Parallels and Lessons
Economic history offers parallels. Early electrification did not immediately boost productivity because factories initially used electric motors as substitutes for steam engines. Productivity surged only after organisational redesign around electricity.
Similarly, in 1987, economist Robert Solow famously observed:
"You can see the computer age everywhere but in the productivity statistics." — Robert Solow
The productivity boom linked to computers arrived only in the late 1990s, after firms adapted structures and processes. AI appears to be following a comparable trajectory.
Technological impact depends on complementary organisational change; without it, productivity gains are delayed.
8. Implications for India’s Growth Strategy
India’s growth strategy increasingly relies on digital public infrastructure, services exports, and the productivity of its large working-age population. AI is already embedded across these domains.
However, its early benefits will manifest as faster workflows, fewer errors, and better coordination rather than immediate GDP acceleration. Policymakers risk underestimating transformation if they rely solely on conventional metrics.
The larger risk is not AI’s failure to transform the economy, but governance systems failing to recognise and adapt to how transformation occurs.
If measurement systems lag behind reality, policy responses may also lag, weakening long-term growth strategy.
Conclusion
AI’s economic impact in India is unfolding quietly through incremental efficiency gains rather than dramatic output jumps. While traditional productivity indicators remain subdued, underlying work processes are changing meaningfully. Updating measurement frameworks and policy interpretation will be essential to accurately assess productivity, guide reforms, and sustain long-term growth in a digital economy.
.jpeg){kind=small}