AI in Healthcare: Opportunities and Challenges

The global application of Artificial Intelligence (AI) in health shows promise in areas like image recognition in radiology and diagnostic analytics, but concerns are arising about its broader implementation, as highlighted during a national consultation on People-led AI in Health in Delhi.

Promise and Limitations of AI in Healthcare

• AI is effective in controlled settings for tasks such as pattern recognition but struggles with real-world application due to the complexity of healthcare, which involves ethical judgments and social contexts.

Concerns about AI Deployment

• Digital Extractivism: Issues of data ownership, derived intelligence benefits, and risks are critical.
• Bias Risk: AI trained on urban data could reinforce socio-economic biases unless grounded in a rights-based framework.

Rights-Based Framework for AI in Health

• Right to Understand: Patients must comprehend their data, demanding AI systems provide clear explanations.
• Right to Local Processing: Health data should be processed locally with explicit consent for cloud sharing.
• Right to Ongoing Control: Continuous control over data and its insights must be maintained by individuals.
• Right to Equity and Access: AI systems require bias audits and should enhance access without creating disparities.

AI's Role versus Human Care

• AI should supplement, not replace, human care, ensuring decisions remain with accountable professionals.

Labour Impact and Public Health Systems

• AI deployment must include labour impact assessments to prevent workforce reductions and ensure health worker dignity.
• AI systems should strengthen public services rather than serving commercial interests.

Conclusion

AI should aid healthcare but not at the expense of human roles. The focus should remain on addressing systemic health challenges through policy, not solely technological fixes. AI must align with public health goals, maintaining health workers and patients at the center.