GPAI Summit and New Delhi Declaration

Table of contents

Context (IE | PIB | TH): India has successfully hosted the annual Global Partnership on Artificial Intelligence (GPAI) Summit for 2023 at the Bharat Mandapam in New Delhi.

The first three GPAI summits were held in Montreal, Paris, and Tokyo.

Global Partnership on Artificial Intelligence (GPAI)

GPAI is a multi-stakeholder initiative bridging the gap between AI theory and practice through research and applied activities on AI priorities.
It will bring experts from industry, civil society, governments, and academia to promote the responsible evolution of AI.
It is built around a shared commitment to the OECD Recommendation on AI.
Initially, it will work on four themes:
1. Responsible AI
2. Data Governance
3. Future of Work
4. Innovation and Commercialisation
Its founding members are Australia, Canada, the European Union, France, Germany, India, Italy, Japan, the Republic of Korea, Mexico, New Zealand, Singapore, Slovenia, the UK and the US.
Launched in 2020 with 15 members, GPAI’s membership now includes 28 countries and the EU.

India became the Council Chair of the GPAI in November 2022 after France.

China, a major tech superpower, is not part of the multilateral grouping.

OECD is an intergovernmental economic organisation with 38 member countries.
It was founded in 1961 to stimulate economic progress and world trade.
India is one of the many non-member economies with which the OECD has working relationships.

In the 2023 summit. GPAI unanimously adopted the New Delhi Declaration, emphasising:
1. Risk mitigation in AI development
2. Equitable resource access for AI innovation
The declaration acknowledges AI’s rapid advancement and recognises its potential for economic growth, innovation, job creation, and societal benefits.
The declaration also flagged concerns about AI systems, including misinformation and bias, unemployment, intellectual property and personal data protection, threats to human rights and democratic values, deepfakes, cybersecurity, and cyber-terrorism.
It also agreed to support AI innovation in agriculture as a new “thematic priority”.
It said that GPAI will aim for diverse membership, prioritising low and middle-income countries.

The world’s first AI Safety Summit was held at Bletchley Park in the UK in 2023.
At the summit, the Bletchley Park Declaration was signed to minimise risks from ‘frontier AI’.
Thus, the centre of the Bletchley Declaration was security and safety risks related to AI.
In contrast, the New Delhi Declaration found a middle path between promoting and regulating AI.
It addresses AI risks but strongly supports innovation in various sectors like agriculture and healthcare.

Frontier AIs are advanced foundation AI models that can pose severe risks to public safety.
Foundation models are trained on diverse unlabeled data, enabling versatile use with minimal fine-tuning for various tasks.
They will replace the task-specific models that now dominate the AI landscape.

India shifted from its view of no AI legal intervention to actively formulating regulations based on a “risk-based, user-harm” approach.
Though the Ministry of Electronics and IT (MeitY) acknowledged AI’s ethical and associated risks, it did not consider any law to regulate the AI sector.
However, after deepfakes of popular personalities gained traction (e.g., actress Rashmika Mandana’s video), MeitY is considering concrete legislative steps to address AI-based misinformation.
At the GPAI Summit, PM flagged AI’s dual potential as a development as well as a destructive tool.
He called for a global framework to provide guardrails and ensure its responsible use.

The declaration supports India’s intention, as Lead Chair for 2024, to endorse efforts to foster collaborative AI for global partnership among GPAI members.
This is a significant win for India, which has advocated a collaborative approach to building AI systems to push its Digital Public Infrastructure (DPI) model worldwide.
Besides, access to member nations’ computing capabilities will also boost India’s plans to build a sovereign AI system.

DPI is a critical digital economy component that enables countries to provide essential services to foster economic growth and social inclusion.
E.g., biometric identity programme Aadhaar and payments solution Unified Payments Interface (UPI).

The government sanctions the underlying technology of components of DPI.
Then, later, it is offered to private entities to develop various use cases.
India wants to take the same DPI approach for developing its sovereign AI.