Public Section Preview
Glossary Terms
| Term (EN) | Definition | Exam Relevance |
|---|---|---|
| Artificial Intelligence (AI) | Machine systems that simulate human cognitive functions — learning, reasoning, decision-making | Core topic term |
| Machine Learning | AI subfield where systems learn from data without being explicitly programmed | AI foundations |
| Algorithm | Step-by-step computational procedure; rules AI follows to reach decisions | Basis of AI |
| Algorithmic Bias | Systematic unfairness in AI outputs due to biased training data | Welfare/policing context |
| Explainability (XAI) | AI system's ability to articulate reasons for its outputs in human-understandable terms | Natural justice link |
| Black-Box AI | AI model whose internal workings cannot be interpreted even by its creators | Due process violation |
| Human-in-the-Loop (HITL) | Framework requiring human review and approval for AI decisions before effect | High-stakes admin |
| Accountability Gap | Diffusion of moral responsibility when AI causes harm across many actors | "Problem of many hands" |
| Predictive Policing | Using AI to forecast crime locations/perpetrators based on historical data | Civil liberties risk |
| Conscience | Moral faculty enabling judgment of right/wrong, integrating reason and empathy | Foundation of ethics |
| Moral Agent | Entity capable of autonomous moral judgment and bearing moral responsibility | Kant's concept |
| Categorical Imperative | Kant's principle: act only according to maxims you would universalise | Kantian ethics |
| Phronesis | Aristotle's "practical wisdom" — ability to deliberate well in complex situations | Virtue ethics |
| Zweckrationalität | Weber: means-end rationality — choosing efficient means to a predetermined end | AI's mode of reasoning |
| Wertrationalität | Weber: value rationality — choosing actions based on ethical values, regardless of efficiency | Conscience's mode |
| Algorithmic Impact Assessment | Pre-deployment audit of AI for bias, fairness, and rights-risks | Canada model, India needed |
| Natural Justice | Legal principle: audi alteram partem (hear both sides) + nemo judex in causa sua | XAI requirement |
| Proportionality | Principle that administrative response must be proportionate to the gravity of the issue | Conscience vs. algorithm |
| Compassion Deficit | AI's inability to feel empathy for vulnerable citizens | Widow pension, drought relief |
| Exclusion Error | AI incorrectly excluding eligible beneficiaries from welfare | DBT, Aadhaar example |
| Responsible AI for All | NITI Aayog's 2021 ethical AI framework for India | India policy |
| India AI Mission | Rs 10,372 crore programme for AI compute, data, safety (2024) | Current affairs |
| Digital Personal Data Protection Act | 2023 Act governing data collection and processing in India | AI data governance |
| Dehumanisation | Treating citizens as data points, stripping them of rights and dignity | AI risk in welfare |
| Veil of Ignorance | Rawls: designing just institutions without knowing one's position in society | Fairness in AI design |
Sources: NITI Aayog, Responsible AI for All (2021); India AI Mission Policy Documents (2024); Max Weber, Economy and Society; Immanuel Kant, Groundwork of the Metaphysics of Morals; Aristotle, Nicomachean Ethics; John Rawls, A Theory of Justice; Luciano Floridi, The Ethics of Artificial Intelligence; Canada Treasury Board Directive on Automated Decision-Making (2019); RPSC Mains Syllabus 2026; ProPublica COMPAS Study (2016).