"AI can democratise healthcare but may also deepen the digital divide." Critically examine this with reference to India's institutional framework for medical AI.

In March 2024, the Union Cabinet approved the IndiaAI Mission with an outlay of ₹10,371.92 crore. Better public service delivery, including healthcare, is one of its goals [1]. Whether AI narrows or widens health inequity depends less on the algorithms than on how institutions govern access, data and accountability.

How AI can democratise healthcare

• Scale where doctors are scarce: The National TB Elimination Programme reported a 27% decline in adverse TB outcomes after adopting AI-enabled tools. India also recorded about 282 million telemedicine consultations between April 2023 and November 2025 [2].
• Shared public infrastructure: Under the IndiaAI–ICMR MoU, ICMR's anonymised datasets go onto AIKosh and researchers get subsidised GPU compute. This lowers entry barriers for startups and researchers [3].
• Ethical baseline: ICMR's 2023 Ethical Guidelines lay down patient-centric principles, an ethics review process and informed-consent norms [4]. AI deployed under IndiaAI must also follow MeitY's AI Governance Guidelines and build in privacy-by-design [2].

How AI may deepen the divide

• Unequal access: WHO lists the accessibility and affordability of the best-performing models as a risk [5]. Well-funded private hospitals can buy better AI, while rural health centres with weak connectivity fall behind.
• Biased data: Models trained on data skewed by race, ethnicity, sex or age [5] may misdiagnose groups that are under-represented in that data, such as tribal and low-income patients.
• Automation bias: Overworked clinicians may miss AI errors [5]. The harm falls hardest on patients who cannot easily get a second opinion.

Gaps in the institutional framework

• Guidelines, not law: No agency is named to assess and approve medical AI before hospitals use it, although WHO recommends one [5].
• No checks after launch: WHO calls for mandatory independent post-release audits [5]. India has none, so privacy-by-design remains a design rule that nobody verifies.
• Unclear liability: If an AI-led diagnosis goes wrong, it is not settled whether the doctor, the hospital or the developer is answerable.

AI's democratising potential is real: it brings scale, shared data and an ethical baseline. Without enforceable safeguards, however, its benefits may go mainly to patients who are already well connected. India should set up a statutory health-AI regulator, require independent audits, test models on diverse Indian datasets and direct IndiaAI funds towards public or not-for-profit infrastructure, as WHO advises [5]. With these steps, medical AI can advance the right to health under Article 21 and SDG 3's goal of universal health coverage.