Democratising AI needs a radically different approach

Context: The growing power and influence of Big Tech companies
is a concern for policymakers worldwide. To break Big Tech’s hold over the
Artificial Intelligence (AI) ecosystem and democratise AI development, India,
like many other countries, is investing in sovereign cloud infrastructure,
creating open data platforms and supporting local start-ups. However, these
efforts are unlikely to be enough and may even deepen Big Tech’s dominance. As
of 2023, Gemini Ultra was the costliest model, costing about $200 million to
train.

Key points

·      
Overview: The Indian government's commitment to advancing
artificial intelligence (AI) technology is evident with its new budgetary
allocation for the IndiaAI Mission.

·      
IndiaAI
Mission: Objective - The mission aims
to establish a robust AI computing infrastructure in India to support the
development and testing of AI systems.

Financial Support - The Union Cabinet approved the Rs
10,372 crore IndiaAI Mission in March to establish a computing capacity of over
10,000 GPUs and develop foundational models with a capacity of more than 100
billion parameters trained on datasets covering major Indian languages for
priority sectors like healthcare, agriculture, and governance.

Current Focus - Initial efforts will involve procuring
300 to 500 GPUs to kickstart the project.

Importance of GPU Procurement - GPUs are critical for
training and building large-scale AI models, essential for advanced AI
applications.

·      
Some
challenges: Limited GPU Capacity and
Infrastructure - The mission's objective to build a high-end AI compute
capacity of 10,000 GPUs is ambitious. Yet, there are concerns about the timely
procurement and deployment of these GPUs to meet the growing demand for AI
applications.

Data Access and Quality - Training AI models on diverse datasets, particularly for Indic
languages, is crucial. However, the current datasets are inadequate for
developing effective indigenous AI models.

Limited AI Expertise and High Costs - There is a shortage of skilled AI professionals in
India. Efforts are being made to address this but bridging this gap remains a
challenge.

High Implementation Costs - The cost of deploying AI solutions, particularly in
sectors like manufacturing, can be prohibitively high. 

Infrastructure Deficiencies - Effective AI deployment requires advanced cloud
computing infrastructure. While efforts like AIRAWAT represent progress, India
still lacks comprehensive AI and cloud computing facilities necessary for
scaling AI applications.

·      
Way
Forward: Incentivize Hardware
Manufacturing - The Production Linked Incentive (PLI) scheme for IT
hardware, notified in 2021, and for semiconductors offers incentives for
increased investment in domestic manufacturing for eligible firms. Expanding
this initiative could further stimulate growth in the sector.

Start-up Support - Provide financial incentives, mentorship, and incubation facilities for
AI startups. Establish AI-focused accelerators and incubators like T- Hub
(India's largest incubation centre) of Telangana.

Comprehensive Data Ecosystem - A National Data Platform can be developed as a
centralised data repository with standardised formats and quality checks and
promote data sharing while ensuring privacy. Invest in rozation and encryption
techniques, as well as data labelling and curation to improve data quality.

Prioritise Ethical AI - Develop comprehensive AI ethics guidelines and regulations, establish
independent AI ethics boards, promote transparency and explainability in AI
systems, and conduct regular AI audits to identify and mitigate biases.

Promote Sustainable AI - Support sustainable AI by investing in energy-efficient AI algorithms
and hardware, promoting the use of renewable energy sources for data centres,
and creating AI-powered solutions for energy optimization and resource
management.