Three Tests for India’s AI Future, Human Capital, Infrastructure, and Institutional Capacity
As global leaders convene in New Delhi for the India AI Impact Summit, a crucial question hangs in the air: Is India ready for the AI revolution? Not just technologically—the country has no shortage of talent or ambition—but structurally, institutionally, and socially.
Economic historians have long demonstrated that technologies spread not simply because they exist, but because societies are prepared for them. Preparation requires skills, infrastructure, institutions, and systems that make technology inclusive. Without these enabling conditions, even the most transformative technologies remain islands of innovation in seas of stagnation.
As technology policy experts Vivan Sharan and Vedika Pandey argue, India’s AI moment must be viewed through this lens. The recent Economic Survey sounded a note of caution, warning that rapid AI deployment could outpace the economy’s structural ability to reabsorb labour. With over seven percent of India’s GDP tied to the IT and IT-enabled services sector, the stakes could not be higher.
Test One: Human Capital and the Compression of the Skills Ladder
Unlike previous industrial transformations that primarily automated manual labour, AI’s impact on jobs is more complex and potentially more profound. AI systems are increasingly capable of writing and improving their own code. They can debug software, generate documentation, and even design algorithms. This capability compresses the ladder of skills that once trained human programmers—the very ladder that India has relied on to build its export strength.
For decades, India’s IT advantage has rested on a vast pool of English-speaking engineers who could be trained in specific technologies and deployed at scale. The model worked because the skills ladder was long and gradual. A fresh graduate could start with basic coding, learn on the job, and gradually ascend to more complex tasks. AI disrupts this progression by automating the lower rungs of the ladder.
The question is not whether AI will displace some jobs—it will. The question is whether India’s workforce can transition to new roles faster than AI eliminates old ones. This is a race between technology and adaptation, and the outcome is far from certain.
But protectionism through delay is a dead end. Slowing AI diffusion to save jobs would inadvertently subsidise domestic inefficiency, leaving Indian industries vulnerable to more agile global competitors who will route around such barriers. Sharan and Pandey draw an illuminating historical parallel: Britain’s attempt to protect its domestic textile industry by banning the export of advanced machinery. British machines were smuggled out and used to establish mills across continental Europe. Britain could not stop the rise of competing manufacturing hubs.
The lesson is clear. The way forward is not to protect workers from AI, but to invest in their capacity to transition. This requires imagining a meaningful social safety net appropriate for the AI era—one that supports retraining, mobility, and income stability during transitions. It also requires encouraging companies to invest in innovation and research and development, rather than resting on existing advantages.
Test Two: Infrastructure and the False Choice Between Cloud and Device
Debates on AI infrastructure often pose a false choice between centralised computing—exemplified by cloud services—and decentralised on-device computing. In reality, developing economies will likely operate somewhere in between, balancing brute force with resilience and access.
The financial dimension of this choice is significant. Cloud computing, once assumed to be infinitely scalable, now faces profitability pressures as AI workloads become vastly more expensive. Training large language models requires massive computational resources, and even inference—running trained models—can be costly at scale.
For India, this creates both challenges and opportunities. The orchestration of AI workloads between the cloud and last-mile devices such as smartphones is a feasible alternative in many use cases. A farmer in rural India does not need a large language model running on a remote server to get crop advice; a well-designed app on a smartphone, with occasional cloud connectivity, may suffice.
This is also where sector-specific deployment becomes crucial. AI is not a single technology but a family of techniques that must be adapted to specific domains. Hospitals need AI that understands medicine. Banks need AI that understands finance. Courts need AI that understands law. Factories need AI that understands manufacturing.
In the past, organisations relied on IT departments to choose software, troubleshoot problems, and train staff on tools. AI is different. It requires in-house expertise that understands both the technology and the domain in which it operates. Building such enlightened workforces may be as important as building data centres.
Test Three: Institutional Capacity and the Risk of Blunt Regulation
The third test is perhaps the most neglected in public discourse. Regulatory and policy institutions that lack the technical capacity to supervise new technologies tend to rely on blunt measures that are often antithetical to progress.
Sharan and Pandey trace this cycle through recent regulatory history. In the face of encrypted communication, legitimate concerns over illicit activity have sometimes led to demands to bypass encryption altogether. This is characteristic of a policy environment where the absence of precise investigative tools leads to “all-or-nothing” approaches. Similarly, in the early days of cryptocurrency, regulatory responses often swung wildly due to limited visibility into its flows. Regulators could not easily distinguish between legitimate and harmful activity, so the default response tilted toward precautionary restrictions.
The unintended consequences were real. India saw digital asset entrepreneurs and innovation migrate to the UAE and Singapore, taking jobs, capital, and expertise with them. The restrictions did not eliminate cryptocurrency activity; they simply drove it offshore and underground.
With AI, the stakes are even higher. If India does not build institutional capacity to understand and respond to AI-driven threats with technical nuance, the country risks a future where online trust is managed through restrictions that constrain innovation, speech, creativity, and the future of work itself.
Regulators, courts, and the executive branch must bridge technical capacity gaps so they are not forced to choose between vulnerability and regressive rulemaking. This means hiring technical experts, investing in training, and engaging with the technology community. It means moving from a posture of suspicion to one of informed oversight.
The Way Forward: An AI Knowledge Consortium
It is in this overarching context that the AI Knowledge Consortium—consisting of 16 research-led institutions—and The Pioneer are hosting a panel discussion on February 19. The conversation will bring together senior tech and policy leaders to examine how AI is reshaping economies, institutions, and societies. It will explore why some economies move from experimentation to widespread use of new technologies while others do not.
The questions are urgent. How can India build the human capital needed for an AI-driven economy? How should it balance cloud and device infrastructure? How can its institutions develop the technical capacity for nuanced regulation? The answers will determine whether AI becomes a force for inclusive growth or a driver of inequality and dislocation.
Conclusion: Beyond the Hype
The India AI Impact Summit is an important moment, but summits are only as valuable as the follow-through they generate. India has the talent, the ambition, and the scale to be a global AI leader. But leadership requires more than declarations; it requires preparation.
The three tests—human capital, infrastructure, and institutional capacity—are not sequential; they are simultaneous. India must invest in all three at once, because progress in one area will be undermined by weakness in another. A workforce ready for AI is of little use if infrastructure is lacking. Good infrastructure is of little use if institutions respond with blunt regulation.
The path forward is difficult but clear. India must embrace AI, not resist it. It must invest in its people, its infrastructure, and its institutions. And it must do so with urgency, because the AI revolution will not wait.
Q&A: Unpacking India’s AI Challenges
Q1: What are the three tests for India’s AI future identified in the article?
A: The three tests are: (1) Human capital readiness, particularly the challenge of AI compressing the skills ladder that once trained human programmers; (2) Access to enabling infrastructure, including the need to balance centralised cloud computing with decentralised on-device computing; and (3) Institutional capacity, especially the ability of regulators and policymakers to understand AI with technical nuance rather than resorting to blunt, precautionary restrictions. These three dimensions must be addressed simultaneously for India to successfully navigate the AI transition.
Q2: How does AI differ from previous technological disruptions in its impact on employment?
A: Previous industrial transformations primarily automated manual labour—tasks that were physical, repetitive, and often dangerous. AI is different because it targets cognitive tasks, including the very skills that have been the foundation of India’s IT advantage. AI systems can now write and improve their own code, compressing the ladder of skills that once trained human programmers. This means the displacement risk is not just for low-skilled workers but for the middle-skilled knowledge workers who have been the engine of India’s services exports.
Q3: What historical parallel do the authors draw regarding protectionism, and what lesson does it offer?
A: The authors draw a parallel with Britain’s attempt to protect its domestic textile industry by banning the export of advanced machinery. Despite the ban, British machines were smuggled out and used to establish mills across continental Europe. Britain could not stop the rise of competing manufacturing hubs. The lesson is that protectionism through delay is futile. Slowing AI diffusion to save jobs will only subsidise domestic inefficiency and leave Indian industries vulnerable to more agile global competitors. The way forward is to invest in workers’ capacity to transition, not to protect them from change.
Q4: Why is the choice between cloud and device infrastructure a “false choice”?
A: The choice is false because developing economies like India will likely operate somewhere in between, balancing brute force with resilience and access. While cloud computing is essential for training large models, many use cases can be handled on last-mile devices like smartphones. Moreover, cloud computing faces profitability pressures as AI workloads become vastly more expensive. The optimal approach is orchestration—distributing workloads between cloud and device based on the specific requirements of each application, sector, and user.
Q5: What risks arise from institutional capacity gaps in regulating AI?
A: When regulators lack technical capacity, they tend to rely on blunt, precautionary measures. The article cites examples from encrypted messaging and cryptocurrency, where limited visibility led to “all-or-nothing” approaches that constrained innovation and drove entrepreneurs offshore. With AI, the stakes are even higher. Without technical nuance, regulators risk imposing restrictions that harm innovation, speech, creativity, and the future of work. Building institutional capacity—hiring experts, investing in training, engaging with the tech community—is essential to avoid this outcome.
