India is rapidly adopting Artificial Intelligence (AI) in government services, recruitment and health care. Yet, people with disabilities are often left out of the design process, leading to inaccessible digital platforms and biased AI outputs.
Key Developments
- In November 2024, the Rajive Raturi judgment directed the Centre to set mandatory accessibility standards within three months.
- The Chief Commissioner for Persons with Disabilities penalised 155 establishments, including government ministries, for non‑accessible websites and apps.
- Research shows AI models exhibit disability bias: 21 language models tested on AccessEval became more error‑prone and stereotypical when disability was part of the query.
- The image model CLIP was 15 percentage points less accurate on photos taken by blind users, and objects like white canes appeared up to 17 times less often in training data.
- A survey of 2,462 NClude platform users revealed that only 543 could complete tasks using AI alone; the rest required human assistance.
- India plans to invest $200 billion in AI over the next two years, while data‑centre capacity is expected to quadruple to **6.5 GW** by 2030.
Important Facts & Figures
- The Rights of Persons with Disabilities Act, 2016 requires accessible digital services, but compliance remains rare.
- Out of 1,313 users who reported completing a previously inaccessible task, only 543 succeeded without any human help.
- Maharashtra reduced its renewable‑energy requirement for data centres from **100 %** to **51 %**, raising concerns about grid stress and heat generation.
- Power‑dependent assistive devices (wheelchairs, oxygen machines) cannot be switched off during grid failures, making power reliability a critical issue for disabled citizens.
Exam Relevance
These developments intersect multiple UPSC syllabus areas: GS2 (rights, Supreme Court judgments, disability law), GS3 (AI funding, data‑centre expansion, energy policy), and GS4 (ethical considerations of bias and accessibility). Understanding these linkages helps answer questions on inclusive growth, digital governance, and sustainable development.
Way Forward
- Mandate disability‑bias testing for every AI system deployed by the government.
- Require that training datasets include genuine, consent‑based disability representation.
- Link state incentives for data‑centres to strict renewable‑energy targets and grid‑impact assessments.
- Establish a permanent advisory body of disabled persons to guide AI policy, data‑centre siting, and emergency‑alert design.
- Enforce the binding standards ordered by the Supreme Court within a clear timeline, with penalties for non‑compliance.