DEEP UPSC EXPLAINEROriginal source-linked study material
The MANAV vision presented AI as a human-centred project rather than a technology whose benefits are automatic. It is a set of policy principles, not proof of enforceable obligations or universal compliance. For UPSC, the central question is how ethical aspirations become practical governance through assigned responsibility, transparent evaluation, accessible services and meaningful ways for affected people to challenge errors.
Key concepts
Human oversight
Meaningful human capacity to review, intervene and remain responsible for consequential use.[2][3]
Contestability
The ability of affected people to question and seek correction of a decision.[2][3]
Ethical impact assessment
Structured examination of possible rights, social and ethical consequences before and during use.[2][3]
Mechanism: MANAV: a human-centred vision for AI
01State public purpose
02Assess rights and risks
03Assign accountable actors
04Test and disclose limits
05Enable review and remedy
Principles become governance through risk assessment, assigned duties, practical testing and accessible remedy.[2][3]
Background and conceptual foundations
International AI-governance frameworks connect innovation with human rights and public accountability. UNESCO’s Recommendation emphasises ethical impact assessment and oversight, while OECD principles address trustworthy AI. Editorial interpretation: ethical language is useful only when it changes decisions about design, procurement and use. A public agency must still decide which tasks are appropriate for automation, whose interests are affected and which safeguards are necessary, rather than treating a declaration of good intentions as sufficient protection.
Editorial model: Principles become operational through a chain of decisions: define the purpose, assess risks, identify responsible actors, test performance and enable review. Transparency supports this process but does not require every user to understand source code. Editorial illustration: a person denied a service needs an understandable explanation and an effective correction route. Publishing a highly technical model description without a usable grievance channel may satisfy a narrow disclosure practice while leaving accountability weak.
Editorial analysis: a human-centred approach can improve public trust and make innovation more durable by addressing foreseeable harms early. Inclusion matters because benefits can otherwise concentrate among people with better connectivity, language access or technical skills. Yet accessibility alone is insufficient if the underlying decision is unreliable. The quality of governance depends on both who can use a system and whether the system is appropriate, robust and contestable in its actual institutional setting.
Editorial analysis: governance must balance innovation, rights and institutional capacity. A rigid requirement applied equally to every low-risk and high-risk use can waste resources, while vague voluntary commitments may leave serious harms unaddressed. Sovereignty also requires careful interpretation: control over infrastructure and data can support autonomy, but isolation may limit learning and competition. The policy task is to identify proportionate safeguards rather than assume that either unrestricted adoption or blanket restriction produces trust.
The February MANAV statement covered ethical systems, accountable governance, national sovereignty, accessible inclusion and valid systems. It should be treated as a policy vision. UNESCO and OECD provide background principles, not evidence that every element has become binding Indian law. Editorial assessment: evaluate subsequent measures separately by their actual legal status, enforcement arrangements and implementation. A new acronym can organise a debate, but it does not itself establish independent oversight or remedy for affected people.
Editorial recommendations: translate principles into procurement conditions, documented responsibilities and proportionate impact assessments. Preserve a responsible human decision-maker for consequential uses and provide accessible review. Test systems across relevant languages and user groups, while publishing meaningful limits and error information. Build public-sector technical capacity so oversight is more than acceptance of vendor assurances. Periodic evaluation should ask whether the system remains useful and justified, including whether a simpler non-AI process would serve the purpose better.
In Mains, distinguish values, governance instruments and outcomes. Use MANAV as a framework to examine a concrete application instead of merely expanding an acronym. In Prelims, avoid treating international recommendations as automatically binding domestic legislation. A strong response explains that accountability requires identifiable duties and remedies; saying that a human remains nominally involved is insufficient if that person lacks authority, time or information to question the system.
Imagine a hypothetical automated scholarship decision where an official can formally override an output but has no explanation or supporting information. Human involvement exists on paper, yet oversight is weak. A useful answer recommends evidence, authority and a realistic review process, demonstrating why human-centred governance must be assessed through institutional practice rather than a label.
