Topic 5: India's AI Policy & The "Builder's Dilemma"

 

Syllabus Mapping

• GS Paper 3: Awareness in the fields of IT, Computers, Robotics, AI; Indigenization of technology and developing new technology.

• GS Paper 2: Government Policies and Interventions for Development in various sectors (Digital Governance).

 

Why in News?

While India has successfully established itself as a leader in AI Deployment (using AI) through its Digital Public Infrastructure (DPI), experts argue it is lagging in AI Creation (building Foundational Models), risking digital sovereignty if it does not address gaps in compute infrastructure and legal clarity.

 

Key Highlights

• The Core Imbalance (Deployment vs. Creation):

  • Success: India excels in deploying AI at scale (e.g., Aadhaar, UPI, DigiLocker, Bhashini) to deliver citizen services. This "DPI Stack" is a global superpower.

  • Failure: India is "shipping applications but not owning the core technology." We are building on top of foreign models rather than building the models themselves, which creates dependency.

• Barriers to "Building" AI in India:

  • Legal Ambiguity: There is no clarity on whether publicly available data can be used to train AI models. The Copyright Act is outdated, leaving developers unsure if they are violating laws by text/data mining.

  • Liability Vacuum: If an AI causes harm, it is unclear who is responsible—the model creator, the deployer, or the platform. This uncertainty deters investment in high-stakes sectors like healthcare.

  • Infrastructure Deficit: Academic institutions lack the "compute backbone" (GPUs). The AIRAWATsupercomputer initiative is criticized for being opaque ("access pathways are unclear") and not fully operational for widespread research.

• Policy Recommendations:

  • Safe Harbours: Create legal safe harbours for developers where liability is proportional and predictable.

  • Regulatory Sandboxes: Establish "structured sandboxes" run by sector regulators to allow informal experimentation before strict rules apply.

 

Critical Analysis 

• Significance (AI Sovereignty):

  • The "Digital Colony" Risk: If India relies solely on US (OpenAI/Google) or Chinese (DeepSeek) models, it surrenders control over the "intelligence layer." Foreign models may not align with Indian cultural or linguistic nuances.

• Challenges:

  • The "DeepSeek" Shock: The article notes that Chinese models like DeepSeek have matched US capabilities at a fraction of the cost, wiping trillion dollars off US tech stocks. This proves that building efficient models is a strategic asset India cannot ignore.

  • Data Privacy vs. Training: The new DPDP Act (Digital Personal Data Protection Act) focuses on privacy but complicates the use of personal data for training AI, creating a conflict between "protection" and "innovation."

 

Value Addition (Web Search - Government Initiative)

• The "IndiaAI Mission" (Budgetary Data): To address the "compute" gap mentioned in the article, the Union Cabinet approved the IndiaAI Mission with an outlay of ₹10,372 crore.

• Specific Component: A key pillar is the "IndiaAI Compute Capacity," which aims to build a scalable AI computing ecosystem of over 10,000 GPUs through public-private partnerships, specifically to help startups and researchers build foundational models.

 

Mains Question

Q.  "India's success in Digital Public Infrastructure (DPI) has created a robust ecosystem for AI deployment, but the country lags in AI creation." Discuss. What policy interventions are needed to transition India from a consumer to a creator of Foundational AI Models? (250 words)

 

Preliminary Question

Q. With reference to the 'IndiaAI Mission' and the AIRAWAT initiative, consider the following statements:

1. The IndiaAI Mission aims to establish a computing infrastructure of over 10,000 GPUs to democratize access for startups and researchers.

2. AIRAWAT is an AI-specific cloud computing platform installed at C-DAC, Pune.

3. The Copyright Act of 1957 currently contains specific provisions granting 'Safe Harbour' to AI developers for training models on publicly available data.

Which of the statements given above are correct? 

(A) 1 and 2 only 

(B) 2 and 3 only 

(C) 1 and 3 only

(D) 1, 2 and 3

Answer: (A) 

 

Explanation:

• Statement 1 is correct: The approved IndiaAI Mission (₹10,372 Cr) targets 10,000+ GPUs.

• Statement 2 is correct: AIRAWAT is India's AI supercomputer installed at C-DAC, Pune.

• Statement 3 is incorrect: The article explicitly states that the Copyright Act has not been updated to address text-and-data mining, and developers currently lack legal clarity or safe harbours.