India’s AI Reforms & the Bharat Rate of Growth: UPSC Mains Notes
In News: AI as India’s 1991 Moment
- For 45 years, India grew at an anaemic 3% “Hindu rate of growth”. A balance-of-payments crisis triggered India’s 1991 liberalisation reforms.
- AI now offers the same transformational leverage that liberalisation provided in 1991.
- India has reached another reform moment to achieve 8% Bharat rate of growth.
Concept: From ‘Hindu Rate of Growth’ to ‘Bharat Rate of Growth’
- Crisis-Reform Lesson: Crisis breeds reform and reform breeds exponential economic outcomes.
- Aadhaar Scale: Aadhaar enrolled 1.38 billion people in the world’s largest biometric system.
- UPI Scale: UPI processes 250 billion annual transactions worth $3.4 trillion globally.
- Jio Disruption: Reliance Jio added 100 million subscribers in just five months in 2016.
- Data Price Fall: Data prices fell from $3 per GB to $0.10 in under three years.
- Regulatory Model: Government created the right environment through spectrum auctions and net neutrality.
Proposal: Free AI Tokens & the R&D Spending Gap
- Low R&D Spending: India spends just 0.65% of GDP on R&D far below global peers.
- Global Comparison: China spends 2.4%, US 3.5%, South Korea 4.9% and Israel 5.4%.
- Token Subsidy Cost: AI token subsidy for top institutions would cost about $2 billion annually.
- GDP Share: This represents only 0.06% of India’s GDP in total expenditure.
- Subsidy Comparison: This is one-fourteenth of food subsidy and one-tenth of fertilizer subsidy.
- Funding Mechanism: Tokens can be funded by freezing subsidy growth for just one year.
- Public-Private Model: India should forge partnerships with AWS, Google and Microsoft for free inference capacity.
Strategy: Sovereign AI, Open Models & Compute Diversification
- Sovereign Hosting: India must build competence to host large language models not merely consume them.
- Example: Through Sarvam, India has shown frontier models can be trained on Indian soil already.
- Open-Source Advantage: Open models offer sovereignty, lower costs and customisation for Indic languages.
- Strategic Capability: AI infrastructure should be treated as a strategic national capability like space programmes.
- Technical Requirements: Hosting LLMs requires multi-region redundancy and 99.99% uptime for critical services.
- NVIDIA Dependence: NVIDIA controls over 80% of AI training hardware creating deep vendor lock-in.
- Hardware Mix: India should adopt a 40:30:30 hardware mix across AWS, Google TPUs and NVIDIA.
Way Forward: National AI Token Policy & DPI Replication
- Policy Timeline: India should implement a National AI Token Policy over the next 24 months.
- First Pilot: Unlimited research tokens should go to top 20 IITs and IISc initially.
- Expansion Phase: Access should expand to 500 startups and 100 universities progressively.
- Literacy Pilot: An AI literacy pilot should launch across 500 high schools in 10 states.
- Full Deployment: The programme should culminate in deployment across 5,000 high schools in all 22 languages.
- Sectoral Application: Fine-tuned models must deploy into healthcare, agriculture, judiciary and education sectors.
Source: The Hindu
