Description

Disclaimer: Copyright infringement not intended.

Context:

India is advancing its ambitions in Artificial Intelligence (AI).
Recently, A step was taken by Union Minister Ashwini Vaishnaw. There was an announcement of establishing an indigenous AI Safety Institute (AISI) under the IndiaAI Mission.
This initiative focuses on safe & responsible development of AI systems.
The AISI will operate under the Safe & Trusted Pillar to address the AI related risks matching with global efforts.

Global AI Safety Initiatives :

Countries like the U.K., U.S., Singapore, & Japan already have set up AI Safety Institutes (AISIs) since 2023.
These institutions aim to manage AI risks through collaboration & research without being hampered by outdated regulations.
Some notable global developments include:
- K. AISI: launched the Inspect platform to evaluate AI models in areas like reasoning & autonomous capabilities.
- S. AISI: Established a taskforce to address national security & safety concerns.
- Singapore’s AISI: Focused on content assurance, safe model design, & rigorous testing.

India’s Approach to AI Safety

India also needs to focus on specific local concerns like AI inaccuracies & the potential for discrimination. It is due to India's diverse social, linguistic, & technological landscape.
India’s AISI will adopt a hub & spoke model collaborating with academic institutions, startups, industries & government bodies to ensure inclusivity and effectiveness.

India-Specific Solutions

Indian startups like Karya are addressing issues like non representative data by helping rural communities. It creates high quality datasets in local languages.
Multilingual AI development: it ensures AI models are accessible in India’s linguistic diversity.
AISI will also prioritise indigenous research & development using local datasets. It will also focus on ethical AI frameworks, risk management, deep fake detection, & other key areas.

Global Collaboration for AI Safety

India's AISI should work closely with the global AISI to understand & mitigate risks like cybersecurity & misinformation.
The Bletchley Declaration signed at the UK AI Security Summit addresses global AI threats, & it also guides the international community in addressing challenges like

Creating Global Standards

For effective governance, India’s AISIs should also align with international standards as well as adapt them for India’s context.
Key steps include:
- Standardised AI safety taxonomy: it should be ensured that experts, policymakers & social scientists are using a common language for AI discussions.
- International notification framework: there should be a system around the world to share information about the impacts of AI models, promote transparency & coordinated governance.
India’s leadership in the global South can enable it to advance inclusive AI governance
Many emerging economies lack the resources to establish their own AISIs. Here India can play a leading role in this direction.
- By developing AI safety frameworks & assessment metrics adapted to local challenges.

Strengthening India’s AI Safety Ecosystem

Collaborations such as the MeitY UNESCO partnership already laid a strong foundation by identifying gaps in AI ethics.
India’s AISI will leverage these information to create frameworks that focus on safe AI development & deployment.
Ongoing projects under the IndiaAI Mission are exploring areas like:
- Machine unlearning (removing harmful data),
- Synthetic data generation,
- AI bias mitigation,
- Privacy-enhancing tools.
These initiatives will form the backbone of a robust AI safety system.

Key Facts about IndiaAI:

● Collaborations:

○ Government of India, NASSCOM, State governments (Telangana, Tamil Nadu).

○ Private companies: IBM, Microsoft, NVIDIA, Meta, Tech Mahindra.

○ Academic institutes: IIT Madras, IIT Jodhpur, IIT Ropar.

○ Startups: CoRover, Niramai, Haptik, Yellow AI, TagHive.

● Government Initiatives:

○ MEITGI (Ministry of Electronics and Information Technology).

○ Digital India and National Programme on AI for promoting AI adoption.

AI infrastructure in india

● IndiaAI Mission (2024): The Indian government has allocated ₹10,300 crore over five years to strengthen the country's AI capabilities.

● AI Compute Facilities: India has 10,000 GPUs (Graphics Processing Units) available for AI computing, with plans to expand further. This is one of the largest AI infrastructures globally.

● Affordable Access to AI: The cost of using GPUs in India is ₹100 per hour, much cheaper than the global rate of $2.5 to $3 per hour.

● Domestic GPU Development: India plans to develop its own GPUs within the next 3 to 5 years.

● Semiconductor Manufacturing: The country is building 5 semiconductor plants to support the AI and tech sectors.

Data & AI Centres:

● IndiaAI Dataset Platform: This platform provides open access to high-quality, anonymized datasets, helping startups and researchers in India.

● Centres of Excellence (CoE): India has CoEs focusing on healthcare, agriculture, and sustainable cities. There's also a new CoE for AI in education with a ₹500 crore funding.

AI Models & Language Technologies:

● Indigenous AI Models: India is focusing on creating its own AI models, especially large language models (LLMs) and AI solutions for specific problems.

● BharatGen: The first government-funded multimodal LLM, aimed at improving public service delivery.

● Sarvam-1: An AI model supporting 10 Indian languages, useful for translation and content generation.

AI & Digital Public Infrastructure (DPI):

● AI is being used in platforms like Aadhaar, UPI, and DigiLocker to enhance digital services.

● For the Mahakumbh 2025, AI will be used for crowd management and providing multilingual services during the world’s largest human gathering.

AI Talent & Workforce Development:

● IndiaAI Future Skills: AI education is being integrated into university curricula, and fellowships are offered for Ph.D. researchers.

● India ranks first globally in AI skill development, according to the Stanford AI Index 2024.

● Since 2016, the number of AI professionals in India has grown by 263%, and the country expects to have 1 million AI professionals by 2026.

AI Adoption & Industry Growth:

● Business AI Investments: 80% of Indian companies are prioritizing AI, with 69% planning to increase their tech investments in 2025.

● GenAI Startup Growth: Funding for AI startups surged six times in Q2FY2025.

● AI in Workplaces: 70% of Indian employees used AI at work in 2024.

● AI for SMBs: 78% of small and medium businesses using AI have reported revenue growth.

AI Startup Ecosystem:

● India has over 520 tech incubators, 42% of which were set up in the last five years.

● AI accelerators like T-Hub MATH are helping startups in product development and scaling.

AI Regulation: The government is taking a balanced approach to AI regulation, encouraging innovation while addressing privacy, cybersecurity, and issues like deepfakes using AI solutions.

Source: ET

Practice Question

Q: Critically examine the significance of establishing the Artificial Intelligence Safety Institute (AISI) under the India-AI Mission. How can it contribute to global AI safety and governance, given India’s unique challenges and its potential leadership in the Global South?

(250 words)