BharatNotes What is Algorithmic Bias?
Algorithmic bias is the tendency of an automated or AI-driven system to produce systematically unfair outcomes for certain individuals or groups — for instance, by gender, race, caste, religion or economic status. The US National Institute of Standards and Technology (NIST Special Publication 1270, 2022) stresses that such bias can arise even without any discriminatory intent, and classifies AI bias into three broad categories: systemic bias (embedded in institutions and society), statistical/computational bias (unrepresentative data and modelling errors), and human bias (cognitive biases of designers and users).
How Does It Arise?
| Source | Mechanism | Verified example |
|---|---|---|
| Biased training data | Model learns historical prejudice present in past records | Amazon scrapped an AI recruiting tool after it penalised résumés containing the word "women's" — it was trained on a decade of male-dominated tech CVs (reported by Reuters, October 2018) |
| Unrepresentative datasets | Under-sampled groups face higher error rates | MIT's Gender Shades study (Buolamwini & Gebru, 2018) found commercial facial-analysis error rates up to 34.7% for darker-skinned women versus 0.8% for lighter-skinned men |
| Flawed proxies and design | Variables correlate with protected attributes | ProPublica's analysis (May 2016) of the COMPAS recidivism tool in Florida found Black defendants were nearly twice as likely as white defendants to be wrongly flagged "high risk" (false-positive rates of about 44.9% vs 23.5%) |
Why It Matters for India
India is deploying AI at population scale — in welfare targeting, fintech credit scoring, facial recognition for policing, and health screening. Biased systems risk amplifying existing inequities of caste, gender, language and region, undermining Articles 14 and 15 values. NITI Aayog's approach paper "Principles for Responsible AI" (February 2021) explicitly lists the Principle of Equality and non-discrimination among its responsible-AI principles. The IndiaAI Mission, approved by the Union Cabinet on 7 March 2024 with an outlay of ₹10,371.92 crore, includes a Safe and Trusted AI pillar supporting responsible-AI projects and indigenous tools.
Current Status of Regulation (as of June 2026)
- India: MeitY released the India AI Governance Guidelines on 5 November 2025 under the IndiaAI Mission — a voluntary, "soft-law" framework built on Seven Sutras including Fairness and Equity, and Accountability; it names bias and discrimination among seven key AI risk areas and asks developers to test for and mitigate bias in training data.
- European Union: The EU AI Act, the world's first comprehensive horizontal AI law, entered into force on 1 August 2024; it follows a four-tier risk-based approach, with high-risk systems (e.g., in law enforcement and critical infrastructure) facing strict data-governance and human-oversight duties, and becomes fully applicable on 2 August 2026.
UPSC Angle
This is a foundational concept that underpins questions on Artificial Intelligence, Responsible/Ethical AI, data governance and emerging-technology regulation in GS3, with crossovers into GS2 (vulnerable sections, transparency in governance) and GS4 (ethics of technology). Aspirants should be able to define the term, cite one verified global case study (COMPAS, Amazon or Gender Shades), name India's institutional responses (NITI Aayog 2021 principles, IndiaAI Mission 2024, AI Governance Guidelines 2025), and suggest remedies — diverse datasets, bias audits, algorithmic impact assessments, explainability and human oversight.
