From Socrates to Gautama Buddha, advances in knowledge have long raised questions of ethics and responsibility. Artificial Intelligence is no exception
AI is no longer a frontier technology. It is becoming part of how businesses and governments operate. As adoption grows, the key question is no longer whether AI can be built, but whether it can be trusted. And how it is being built.
This shift, from building AI to trusting it, is bringing ethics into how systems are designed, deployed and used. It also asks us to rethink what we mean by engineering. Building trustworthy AI is not just a technical problem. It is equally about context, judgment and responsibility.
The real task is not just to define principles, but to put them into practice and to shape people and institutions that will ultimately build and govern these systems.
Much of the early discussion on AI ethics focused on bias, opacity and accountability. These issues remain unresolved. But the scope of challenges is widening as AI systems beginning to influence critical decisions in areas of real consequence; credit, healthcare, hiring and public services.
The risks are no longer occasional and sporadic. They are systemic and increasingly part of everyday economic and social decisions.
Early use of AI in digital lending shows that borrowers with limited formal credit history can be disadvantaged not by intent, but because proxy variables reflect income volatility or informal work patterns. These are not failures of intent; they are failures of design.
From principles to capability
India’s context magnifies both the opportunity and the risks. Its scale and diversity amplify outcomes, while its digital public infrastructure allows AI to operate at unprecedented scale. India will not just be a large market for AI, it is also a real-world test bed where the challenge of building responsible AI at scale will unfold in full view.
This creates a unique opportunity. India can define a middle path supporting innovation while building trust.
But this will require moving beyond high-level principles. Many organisations today have AI ethics statements. In practice, a clear gap remains. Ethics is treated as a checkpoint, perhaps an afterthought, not as part of design.
We have seen this before. In manufacturing, quality was once inspected at the end of the assembly line. Over time, it was built into design and production. That shift made companies more competitive.
AI needs a similar shift, from principles to capability, from review to design. Trust has to be built into systems from the start.
This also requires organisations to rethink how AI systems are measured and managed. Traditional metrics — accuracy, efficiency, speed — are no longer enough. Systems must also be assessed for fairness and robustness, particularly in high impact use cases.
Equally important is the need to make AI systems auditable over time. As models learn from new data, their behaviour can change in ways that are not always visible. Continuous monitoring, periodic review and clear escalation mechanisms become essential to maintaining trust.
Finally, leadership attention matters. Trust in AI cannot be left only to technical teams. It requires board-level ownership, much like financial risk or cybersecurity. When organisations treat trust as a strategic priority, it begins to shape both design choices and deployment decisions.
In practice, this means testing for bias, documenting assumptions, making accountability clear and building systems that are robust and auditable. It also means creating teams that can challenge and stress-test systems before deployment.
Shaping responsible AI builders
Institutions beyond industry have a critical role.
Academia must bridge disciplines. AI ethics cannot be addressed by engineering alone. It needs inputs from philosophy, social sciences, law and public policy. This calls for a new kind of talent; engineers who understand both systems and society.
Institutions such as the IITs are well placed to lead this shift. A key but often overlooked factor is the role of the humanities and liberal arts. Technical education equips engineers to optimise systems. It does not always prepare them to interrogate the context. The humanities help build three capabilities that are fast becoming indispensable.
First, contextual understanding: the ability to see that data accurately reflects society.
Second, moral imagination: the ability to anticipate unintended consequences.
Third, judgment under ambiguity: the ability to take decisions when trade-offs are not clear.
This is not theoretical. In several early digital systems, including in finance and welfare, solutions that looked efficient in design created exclusion in practice. This kind of thinking cannot be optional. It is central to building trustworthy systems that work in the real world.
This approach is already beginning to take shape. For instance, AI-driven digital forensics developed at IIT Kanpur has been deployed in public systems to detect fraud while maintaining fairness, showing how technical capability and ethical judgment can be designed together, not added later.
Government too, has a role. It must strike a balance. Too much regulation can slow innovation. Too little can erode trust. The focus should be on clear but flexible frameworks. Building trustworthy AI will also require continuous feedback from society, where systems improve through real-world use and scrutiny.
Trust as the real currency
India’s experience with digital public infrastructure offers a useful lesson. Success was achieved with the interplay of open standards, institutional trust and iterative policies.
The same approach can guide AI.
The cost of missteps is steep. When trust is lost, adoption slows, regulation tightens and momentum slows. It is hard to regain trust, once lost.
As AI adoption accelerates, true constraint will not be technology, but trust. Systems that are demonstrably fair, transparent and accountable will scale. Those that are not will meet resistance.
Trust is more than a societal requirement. It is an economic enabler that will shape adoption, determine scale and define competitiveness. India has the capacity to lead but only if it invests in trust as deliberately as it is investing in technology.
The next phase of AI will not be defined by who builds the most powerful systems, but by systems that people are willing to trust and confidently use at scale.
Sondhi is Independent Director: Global – India, former MD & CEO, Ashok Leyland and JCB India; Verma is Professor, IIT Kanpur, Former Secretary, Science and Engineering Research Board (SERB), Government of India. Views are personal
Published on June 23, 2026
