Perspective
The present era is one of turbulence, a whirlwind of ideas, information, invention, and innovation, accompanied by explosive growth in science-based technology. It is an era where all elements are critically in flux. Viewed through a proper historical and comparative lens, the confluence of innovation, big data, artificial intelligence (AI), Machine Learning (ML), Deep Learning (DL), robotics, automation, cloud computing, advanced data analytics, digitalisation, the Internet and entrepreneurship transformed the rules of the game and introduced a “new normal” in this VUCA (volatility, uncertainty, complexity, and ambiguity) world- a world characterised by disruptive innovations and gaps in regulation and certification, with rising compliance and technological risks across industries. The world of dramatic, blistering change reminds me of a centuries-old Chinese saying, “May you live in interesting times”!
Emerging fields, such as quantum computing, force us to think beyond today’s capabilities. In this explosion of information and data, I sometimes wonder, like these prescient words, “Where is the wisdom we have lost in knowledge? Where is the knowledge we have lost in information” (T.S. Eliot)? This progression from information to knowledge and then from knowledge to wisdom is sorely missing today. In this VUCA world, a world of disruptive innovations and stricter regulatory compliance, it is not enough to visualise what these technologies might eventually accomplish, but how we should decide which developments deserve our attention, investment and trust. Hence, there is a manifest need to balance culture, process, and people.
For organisations, knowing when to adopt a new technology can be difficult. Waiting too long could give competitors a head start or drive customers and talented employees elsewhere. But rushing in can tie up money and people in systems that are not fully understood, properly secured or easily replaced. Agility means recognising changing conditions and bringing together the right people, skills and resources to respond in time. ‘Agile sprinting’- agile in sales and products, sprinting in technology, agile in architecture- combines adaptability, flexibility, innovation and responsiveness to deliver cost-effective, quality-driven solutions to meet evolving customer demands. Ultimately, agility strengthens an organisation’s competitiveness and resilience in a dynamic business environment. The challenge is, therefore, to develop the judgement to distinguish useful progress from expensive distraction. Gary Hamel stressed, “Industry revolutionaries don’t tinker at the margins; they blow up old business models and create new ones”. As I have repeatedly demonstrated in my writings and speeches, there must ultimately be a sequential move from product innovation to lowering costs and finally to creating unique value.
AI’s transformative potential is evident across sectors. In healthcare, AI will not replace doctors, but doctors who use AI will outperform those who do not. By augmenting human expertise, it enables earlier diagnosis, personalised treatment and better outcomes. In education, AI can make world-class learning accessible to every child, replacing one-size-fits-all approaches with personalised, inclusive experiences. It is also reshaping how organisations communicate, manage information, engage stakeholders and demonstrate impact. More broadly, AI is boosting innovation and productivity, addressing complex global challenges and expanding human capabilities. Its greatest promise lies not in replacing human potential, but in enhancing it and redefining the boundaries of human achievement.
Recent reporting in The Economist has highlighted both the difficulty businesses face in extracting measurable returns from AI and the argument that ambitious infrastructure investment could accelerate adoption and improve living standards. These perspectives are compatible: a technology can have substantial long-term potential while individual investments remain uncertain or unsuccessful.
Granular Examination
A company may launch numerous AI initiatives without improving customer experience, profitability or resilience. In “Welcome to the AI trough of disillusionment”, The Economist (May 2025) described the gap between enthusiasm for generative AI and the practical difficulties of deploying it. Citing S&P Global, it reported that the proportion of surveyed companies abandoning most of their generative-AI pilot projects rose from 17% to 42%. Obstacles included fragmented data, older IT systems, shortages of specialist talent and concerns about damaging mistakes.

Innovation Requires Financial Discipline
Technology choices belong in strategic planning and capital allocation, not siloed inside the IT department. The true financial impact goes far beyond initial purchase prices or software subscriptions. A realistic business case must account for the full lifecycle comprising integration and data preparation, cybersecurity and ongoing maintenance, employee training and human oversight, and eventual system replacement.
As The Economist noted in its August 2026 piece, “How to measure returns on AI,” companies must distinguish between usage and actual value. Tracking active licenses, prompt counts, or compute spend only shows that people are using the tool; not that it is improving performance. Real return on investment shows up in team productivity, work quality, and customer satisfaction, even if early adoption causes temporary friction. Shaving hours off a task means little if staff must spend that saved time fixing AI errors. Likewise, generating twice as many reports is not progress if the underlying analysis is weaker, or if leaders simply drown in the volume.
Bottom line
For the unversed, experimentation has its place. Testing an idea on a small scale can help an organisation decide whether it is worth a larger investment. But a trial needs a clear purpose, a way to judge the results and a point at which to decide whether to continue. The first question should be: what problem are we trying to solve? Reducing errors in invoice processing, planning maintenance more effectively or helping staff find reliable information are concrete goals. They give an organisation a much better reason to invest than a vague instruction to “become AI-driven”.
Organisations should establish a baseline before deployment and compare subsequent results against it. They should also ask what happens to any time released: is it used to improve service, increase output, reduce backlogs or undertake more valuable work? Financial prudence should not mean demanding immediate returns from all experiments but knowing what is being tested, how much learning is worth and when continued expenditure requires stronger evidence.
Obsolescence, Flexibility and the Cost of Dependence
Rapid innovation creates an uncomfortable investment problem: a system can remain technically functional while becoming commercially unattractive. A newer alternative may offer lower costs, better performance or stronger integration. Conversely, replacing a functioning system too frequently can itself destroy value. The appropriate response is to build flexibility into investment decisions.
Before committing to a platform, organisations should examine whether they can export their data, connect the system to other applications and move to another provider without prohibitive disruption. Contractual exit arrangements deserve as much attention as introductory pricing. A low initial price may be poor value if subsequent migration becomes costly or impractical. Useful questions include: Can the investment be introduced sequentially? Which components can be upgraded independently? What happens if the supplier changes prices or withdraws support? Who owns the data and any customised outputs? What would an orderly exit cost? These are governance recommendations rather than predictions about which technology will win. Their purpose is to preserve room to adapt. Under uncertainty, retaining options can be more valuable than making the largest possible commitment at the earliest opportunity.
Reading between the lines – AI Boom
The scale of investment surrounding AI makes financial literacy important. Leaders need to understand not only the technology they are buying, but also the assumptions supporting their suppliers’ promises. In its October 2026 article on unreliable numbers in the AI boom, The Economist questioned the clarity and comparability of headline financial measures. It noted that annualised revenue figures can be extrapolated from a short period and that companies may treat sales through distribution partners differently. Such measures should not automatically be read as equivalent to recognised annual revenue or dependable recurring income. Accordingly, a big number means something if you know what’s behind it. Still, healthy scepticism shouldn’t turn into automatic pessimism. In its 2026 leader, “Nvidia is driving the AI boom. Good,” The Economist defended bold risk-taking, pointing out that smart investment can speed up AI adoption and boost productivity, even while acknowledging how uncertain that timeline and the long-term value of computing hardware really are. Put together, this points to a sensible middle ground. Innovation means betting on uncertain outcomes, which is more reason to insist on clear assumptions, honest accounting, and knowing who takes the hit if things go wrong.
Data Centres
In my interviews with Nikkei Asia and India Today, I highlighted growing local opposition to data centres, stemming not merely from concerns about their construction but also from deeper scepticism about the technological requirements. Unlike hospitals or other public infrastructure, which generate visible and immediate local benefits through employment and ancillary economic activity, data centres offer relatively limited direct advantages to host communities. The asymmetry between widely dispersed benefits and concentrated local costs causes concern. While businesses and users elsewhere reap the advantages, communities bear the immediate pressures on land, water and energy resources, leading to NIMBY (Not in My Backyard) resistance.
Tangible socio-economic values of AI are uncertain. Despite the hype surrounding Large Language Models (LLMs), their contribution to productivity, employment and everyday welfare remains unestablished. They can write books, compose music and assist with routine tasks, but whether these capabilities translate into commensurate public value is unsettled, necessitating building public trust through transparency and meaningful community engagement. Clear and enforceable commitments on water and energy consumption, alongside advance disclosure of the environmental and economic costs, should constitute an integral part of project approvals.
Technological ambition cannot substitute for social legitimacy. Data centres and AI infrastructure can be built. But it must be examined whether their wider economic and technological benefits justify their local environmental and social costs and a meaningful say for the affected communities in the decision-making process.
Cybersecurity – Organisational Resilience
Digital transformation also requires a different attitude towards security. When essential processes depend on interconnected systems and external providers, cybersecurity is inseparable from business continuity. IBM’s 2025 Cost of a Data Breach Report placed the average total organisational cost of a breach in India at ₹22 crore. Its 2026 report subsequently put the figure at ₹25.5 crore, an increase of 15.9%. These are study averages, not predictions of what every organisation will lose, but they demonstrate the potential financial significance of a serious incident. The 2026 research, conducted by the Ponemon Institute and sponsored and analysed by IBM, covered breaches experienced by 602 organisations globally between March 2025 and February 2026. Its findings should therefore be understood within that research framework, rather than treated as a complete census of cyber incidents.

For leadership, the practical question is whether the organisation can continue operating when prevention fails. In such crisis situations, Disaster Management, Business Continuity Planning (BCP), Supply Chain, and quick recovery from disruption assume great significance. Boards should seek evidence that critical data is identified, access is controlled, backups can be restored and incident-response responsibilities are understood. Recovery exercises should test real decisions: who can shut down a compromised service, who contacts customers and how essential operations continue. Security becomes meaningful when it is demonstrated, not just declared.
AI’s Promise and Trust Boundaries
AI’s attraction lies in its ability to assist with information-intensive work. Yet useful assistance does not justify unconditional trust. The Economist’s 2026 reporting on AI agents highlighted concerns about plausible but difficult-to-detect errors, data leakage and unauthorised tool use—risks that become especially important when systems undertake longer sequences of actions. A responsible deployment policy should distinguish between generating a draft and exercising authority. Consider a hypothetical progression. Think about how an AI customer service tool gets rolled out: first it summarises incoming emails, then it drafts replies, and eventually it gets the green light to issue refunds or change account details. Each step raises the stakes. Just because the AI is great at summarising doesn’t mean it’s ready to handle money or sensitive data. That’s why oversight should match how serious—and how reversible—a mistake would be. Drafting a low-stakes email leaves plenty of room to experiment. Making calls on jobs, credit, healthcare, or essential services demands much tighter review, clear paper trails, and a way for people to push back. Simply put: the bigger the decision, the stronger the proof, accountability, and guardrails need to be.
Meaningful Human Oversight
“Human in the loop” sounds reassuring, yet it often masks a hollow process. Someone who lacks the time, expertise, or authority to question an AI’s recommendation is just rubber-stamping it. Real oversight requires Context: Enough expertise to evaluate the AI’s output; Visibility: Access to the underlying data driving the decision; and Power: Genuine authority to override the system.
Organisations must also name a clear owner for every AI-enabled process. When something breaks, accountability shouldn’t evaporate between the vendor, the engineering team, and the business unit. Vendor contracts can divide legal obligations, but internal governance still demands a single, named decision-maker. For high-stakes systems, organisations need concrete guardrails, viz., documented testing before deployment, strict limits on what the AI can approve autonomously, audit trails of major decisions and clear escalation paths, and a reliable kill switch to pause the system immediately.
As The Economist has noted regarding agentic AI, these controls prevent silent errors and unauthorised actions. Ultimately, the goal is to ensure a human understands what the system is doing, can challenge its logic, and can stop it in its tracks. Oversight means nothing if the person in charge cannot change the outcome.
Workforce Transition Transcending Headcount
Technology strategy is also a people strategy. Discussions that focus exclusively on jobs eliminated miss the larger challenge of changing tasks, developing capabilities and helping people move between roles. The World Economic Forum’s Future of Jobs Report, 2025 estimates that 39% of workers’ existing skill sets will be transformed or become outdated between 2025 and 2030. It also projects 170 million jobs created and 92 million displaced, producing a net increase of 78 million. Importantly, these are employer-informed projections reflecting multiple structural forces—including technology, demographic change, economic conditions and the green transition—not AI alone.
A positive global balance would not remove the difficulty facing an individual whose role disappears. New opportunities may require different skills, arise elsewhere or offer different working conditions. Organisations should therefore assess which tasks can be automated, which should be augmented and which capabilities must be preserved. Training should form part of the investment plan from the outset, rather than appear as a corrective measure after deployment – training should form a Chapter in the transformative saga rather than a footnote. The Economist’s analysis of AI returns also cautions against treating time savings simply as a call to reduce headcount. It discusses alternatives such as redeployment and avoiding additional recruitment as output grows. Clearly, then, productivity can foster expansion and better service, not just cost reduction.

Harder Work -Organisational Change
I strongly argued in my Keynote Address on Digital Innovation and Sustainable Transformation at Maharishi Markandeshwar (Deemed University), Ambala on 30 July 2026, that making a tool useful is an organisational undertaking. The difficulties identified in The Economist’s reporting—siloed information, legacy systems and shortages of expertise—show why even capable technology may struggle inside an unprepared organisation. Its analysis of returns also describes a productivity “J-curve”: performance may initially decline while people learn and processes adjust, before benefits emerge. This creates a difficult management responsibility. Leaders must allow enough time for learning without permitting indefinite underperformance.

This requires giving each major project a clear business owner, defining what success looks like, and evaluating progress regularly. Tech teams need to be honest about what a system can and can’t do, while operations teams explain how the work gets done day to day. Most importantly, the people closest to the work should help spot the messy edge cases that neat software demos almost always miss. Good change management also means acknowledging people’s anxiety. Nobody is going to speak up honestly if every conversation about “efficiency” feels like a disguised layoff exercise. The goal is to build a culture where people can flag bugs, challenge assumptions, and suggest better ways of working without getting labelled as “anti-progress.” I have always believed that honest pushback isn’t a roadblock to rollout; it’s how I made it work.
Technologically Informed, Not Necessarily Technically Expert
I do not consider myself technology-savvy, and I would not pretend to possess the expertise required to evaluate every emerging system independently. The range of developments is too broad, and the pace too demanding, for superficial familiarity to substitute for specialist knowledge (a specialist knows increasingly more about the less). But this limitation should encourage informed engagement, not withdrawal. There is an important distinction between being technologically informed and being technically proficient. A leader need not write code or design an AI model to ask whether a proposal serves a genuine need, whether its benefits are measurable and whether its risks are acceptable. The responsibility is to know enough to ask useful questions and to recognise when an answer is incomplete. Structured learning through regular briefings, practical demonstrations, independent expert advice and discussions that connect technical choices with financial and operational consequences can help. A supplier’s explanation may be valuable, but it should not be mistaken for an independent assessment. Leaders do not have all the answers – no one does. But they should be willing to ask questions rather than accept a proposal they do not fully understand. Sometimes that means saying, “I need to understand this better. What assumptions are we making, what other options do we have, and what could happen if we go ahead?” Asking the right questions is often the most important – the classic example being Newton’s question of why an apple falls to the ground. In my writings over four decades, I repeatedly asked three basic questions -what, how and why. To this list, I usually add: what now? How can things be changed and changed for the better?
Practical Agenda for Informed Oversight
Approval of a significant technology initiative should be contingent on a satisfactory response to these seven questions: 1. What organisational problem are we solving, and why does it matter? 2. What evidence supports this solution over simpler alternatives? 3. What is the full cost, including implementation, oversight and exit? 4. How will we measure improvements in quality, productivity or service? 5. What could go wrong, and how would we recover? 6. Who remains accountable, and which actions require human approval? 7. What skills and capabilities must we develop to remain adaptable? These questions will make uncertainty visible and decisions more deliberate.
Keeping Technology in Human Hands
Real progress isn’t about the technology itself; it’s about what it empowers people and organisations to do. When it’s working, it leads to better services, smarter workflows, and lasting value. “Technology is a good servant but a bad master” remains a useful warning, provided it is translated into practical choices. Remaining in control means understanding dependencies, retaining the ability to intervene and accepting responsibility for decisions made with technological assistance. We cannot be experts in every new technology, but we should know enough to question what it promises, understand where it could be useful and keep track of how it is used. Adopting the latest system must lead to delivery of real benefits at an acceptable cost and risk, with people still responsible for the decisions made and the actions taken.
NDTV’s Summit: Key Highlights
AI is rapidly reshaping economies, businesses and societies, creating unprecedented opportunities alongside concerns about employment, ethics, security and human autonomy. NDTV’s two significant initiatives in 2026—the NDTV IND.AI Summit held in New Delhi on 18 February and the NDTV World Summit on 22–23 October—provided an important platform to examine these developments. Held under the overarching theme “A World Between Orders,” the NDTV World Summit brought together political leaders, technology experts, entrepreneurs and researchers to discuss AI’s implications for economic growth, employment, national security and society. A major session, “AI at Scale: Building the Intelligence Economy,” featured Andrew Ng, a renowned AI researcher, founder of DeepLearning.AI and co-founder of Coursera. Other discussions explored AI, human consciousness, warfare and the governance of emerging technologies.
AI and the Future of Employment
Andrew Ng cautioned against exaggerated fears of an employment catastrophe, emphasising that continuous learning, upskilling and reskilling would help workers remain relevant. Software engineering, among other professions, is already experiencing significant change. The larger challenge is to use AI to augment human capabilities, improve productivity and create new opportunities rather than merely replace workers.
Economic Growth and Commercial Viability
AI offers considerable potential to automate routine tasks, improve decision-making, enhance customer service and raise organisational efficiency. However, Nikhil Kamath drew attention to the substantial investments flowing into AI and questioned whether commercial revenues would justify the valuations and expectations. This raises an important distinction between technological excitement and demonstrable economic value. AI adoption must ultimately deliver measurable productivity gains and sustainable returns.
Ethics, Human Values and Security
Discussions involving Swami Sarvapriyananda raised deeper questions about intelligence, creativity, consciousness and the distinctive attributes of human beings. Concerns about misinformation, algorithmic bias, privacy, accountability and human oversight underline the need for responsible AI development. Cybersecurity and national security are of paramount importance. While AI can strengthen intelligence gathering and cyber defence, it can also facilitate sophisticated fraud, disinformation and cyberattacks, necessitating appropriate safeguards and international cooperation.
Way Forward
Several priorities emerge from these discussions:
- Strengthen AI literacy, digital education and lifelong learning.
- Promote indigenous research, computing capacity and Indian-language AI applications.
- Ensure responsible innovation through transparency, safety and accountability.
- Support workers affected by technological disruption through reskilling.
- Develop proportionate regulation that protects society without stifling innovation.
- Evaluate AI investments through measurable productivity, accessibility and long-term economic returns.
The overarching message was that AI is not merely another technological advancement; it is a transformative economic and societal force. Its success must be judged not simply by computational sophistication or investment volumes, but by its contribution to productivity, human welfare and inclusive development. For India, the opportunity is substantial, but it necessitates sustained investment in people, research and infrastructure, supported by sound governance. Ultimately, AI must complement human intelligence, strengthen human capabilities and serve society. Evidently, it is a means to an end rather than an end.
“The Times They Are A-Changin’” (Bob Dylan, 1964)-Resetting Strategic Priorities
India’s unparalleled “digitalisation-driven productivity gains” (IMF) have been fuelled by government initiatives, technological advancements, and the digital wave, from building extensive digital public infrastructure and transforming banking and governance to deploying advanced AI, analytics, and cloud at an enterprise scale. This inexorable process is expected to continue, with a billion-strong digital population, visionary public platforms (India Stack, UPI, DigiLocker), and expanding AI and fintech ecosystems.
India is well positioned to benefit from AI through its technology workforce, entrepreneurial ecosystem and expanding digital infrastructure. Agriculture, healthcare, education, finance, manufacturing and public administration offer significant possibilities. The proof of the pudding lies, however, in the eating. As I have stressed, realising this potential requires greater investment in research, computing infrastructure, semiconductor capabilities, technical education and AI solutions suited to India’s socio-linguistic diversity.
A strategic roadmap focusing on infrastructure expansion, domestic cloud and AI models, ecosystem innovation, training, and strong governance can help India become not only a digital adopter but a leader in digital transformation and sovereign digital technology. With the right investments in people, infrastructure, ethical AI, cybersecurity, and inclusion, India can become a model digital economy—inclusive, resilient, self-reliant, and globally competitive. Let us move full steam ahead without ignoring the pitfalls to realise India’s strategic opportunity. This thought is succinctly captured in these Urdu lines,
“ज़िंदगी के अजब ही अफ़साने हैं यहाँ,
तीर भी चलाने हैं और परिंदे भी बचाने हैं।”
Translation: Life has its own strange stories; here, we must fire arrows and yet save the birds also.
SAP’s November 2026 report shows that India is on track to cross the $5 trillion economic milestone in the near term and become the world’s third-largest economy, while positioning itself as a global centre for technology, innovation and AI. India’s large-scale data generation, a deep developer ecosystem, growing investments in AI technologies and an evolving regulatory framework help the country shape how AI and AI agents are built, deployed, and scaled globally.
However, challenges of the digital divide, talent shortages, cybersecurity and privacy risks, and dependence on foreign technology remain. Achieving the goal of Viksit Bharat by 2047 requires innovation transcending individual organisations. It must raise productivity, create opportunities for entrepreneurs and help more people build the knowledge and skills needed to participate in a growing economy. This necessitates policies and programmes that translate ambition into practical support—strengthening existing capabilities while opening new avenues for growth to meet the “revolution of rising expectations”. As Ram Dhari Singh Dinkar wrote eloquently,
“ये नखत अमा के बुझते हैं”
सारा आकाश तुम्हारा है।
Translation: The stars of the moonless night are fading;
The entire heavens are yours to claim.
Tathastu (Sanskrit, meaning “So be it”)!
ABOUT THE AUTHOR
Dr. Manoranjan Sharma has been the Chief Economist for 25 years with Canara Bank and Infomerics Ratings.
Globally recognised as an expert in global economy, Indian economy, banking, finance, MSMEs, sustainable development and financial inclusion, he has chaired sessions at national and international conferences and is recognised as one of the 30 global experts by the UN Virtual University (UNVU) for SDGs.
He has been on over a dozen Advisory Councils in seven states, viz. Member, Rajasthan Governor’s Advisory Council; Governor’s Senate Nominee on Rajasthan University; Governor’s Senate Nominee on SNDT University, Mumbai; and Expert on selection panels for IIMs, Vijay Patil School of Management, SAIL, IDBI Bank, Canara Bank, Bank of India, Union Bank of India.
He has been published widely and quoted extensively in major Indian and international media and publications.
Smt. Archana Choudhary, IRS (Retd.), superannuated in October 2024 as Principal Director General of Income Tax, where she championed reforms for better citizen trust and mitigation of public grievances. A recipient of the Gold Medal for Academic Excellence (1987) at the Lal Bahadur Shastri, National Academy of Administration, she is passionate about women’s empowerment.



