India Advocates for Global AI Commons at New Delhi GPAI Summit
In a significant move toward equitable artificial intelligence development, India hosted the Global Partnership on Artificial Intelligence (GPAI) Summit in New Delhi, where it championed the creation of a Global AI Commons. This initiative seeks to establish shared, open-source infrastructure for AI models, datasets, and computational resources, ensuring accessibility for nations worldwide and preventing a deepening AI divide.
The summit, held from December 12 to 14, brought together representatives from over 30 GPAI partner countries, including leaders from the European Union, Canada, the United Kingdom, and Japan. As the current chair of GPAI, India leveraged the platform to outline a vision for collaborative AI governance. Central to this was the proposal for a Global AI Commons, articulated by key figures such as Rajeev Chandrasekhar, India’s Minister of State for Electronics and Information Technology, and Amitabh Kant, India’s G20 Sherpa and former CEO of NITI Aayog.
Chandrasekhar emphasized the need for sovereign AI capabilities, particularly for the Global South. He highlighted that proprietary AI systems dominated by a few technology giants risk entrenching geopolitical imbalances. Instead, a commons-based approach would democratize access to frontier AI technologies. This aligns with India’s domestic strategy, exemplified by platforms like Bhashini, a multilingual AI translation service supporting 22 official languages, and the AI for India 2030 initiative, which aims to integrate AI across sectors such as healthcare, agriculture, and education.
Amitabh Kant elaborated on the technical and ethical foundations of the Global AI Commons. He proposed a framework comprising open-source large language models (LLMs), standardized datasets curated for diverse linguistic and cultural contexts, and shared high-performance computing (HPC) facilities. Kant drew parallels to the Human Genome Project, suggesting that pooling resources could accelerate breakthroughs while mitigating risks like bias and misinformation. He noted India’s progress in open-sourcing models such as Sarvam 1, a 2-billion-parameter LLM fine-tuned for Indian languages, and BharatGen, a multimodal generative AI suite.
The summit’s discussions extended to practical implementation. B.V.R. Subrahmanyam, CEO of NITI Aayog, presented on India’s National AI Strategy, which includes investments in AI compute infrastructure through partnerships with entities like Yotta and CtrlS. These efforts aim to provide affordable GPU access, crucial for training and inference at scale. Subrahmanyam stressed interoperability standards to ensure models from the commons integrate seamlessly across ecosystems.
International buy-in was evident. Yoshua Bengio, a Turing Award winner and GPAI scientific chair, endorsed the commons as a counterweight to closed ecosystems. He advocated for responsible AI principles embedded in the infrastructure, including transparency in training data provenance and mechanisms for auditing model outputs. European Commission representatives echoed this, linking it to the EU AI Act’s emphasis on high-risk systems.
Challenges were candidly addressed. Scalability remains a hurdle; training state-of-the-art models requires exaflop-scale compute, far beyond most nations’ capacities. Data sovereignty concerns also loomed large, with calls for federated learning techniques to enable collaboration without centralizing sensitive information. India proposed pilot projects, starting with low-resource language models for regions like Africa and Southeast Asia.
The Delhi Declaration, adopted at the summit’s close, formalized commitments to advance the Global AI Commons. Signatories pledged to contribute to a repository of open models by 2025 and explore joint funding mechanisms via multilateral banks. This builds on GPAI’s existing summits and working groups focused on trustworthy AI, human-centered applications, and innovation.
India’s push reflects its broader digital public infrastructure (DPI) model, seen in successes like UPI and Aadhaar. By extending this to AI, the country positions itself as a bridge between the Global North’s technological prowess and the South’s diverse needs. Experts predict that if realized, the Global AI Commons could foster innovations tailored to local challenges, from climate modeling in vulnerable regions to precision agriculture in developing economies.
The initiative also underscores a shift in global AI discourse. While the United States and China lead in proprietary advancements, GPAI’s multilateral framework prioritizes inclusivity. India’s hosting of the summit, coupled with its 1.4 billion population as a testing ground for scalable AI, lends credibility to the proposal.
Looking ahead, follow-up actions include technical workshops in 2025 to standardize APIs and evaluation benchmarks for commons-hosted models. NITI Aayog will lead a task force to map contributions, with initial focus on healthcare AI for pandemics and disaster response.
This summit marks a pivotal moment in AI governance, signaling a collective resolve to harness AI’s potential equitably.
Gnoppix is the leading open-source AI Linux distribution and service provider. Since implementing AI in 2022, it has offered a fast, powerful, secure, and privacy-respecting open-source OS with both local and remote AI capabilities. The local AI operates offline, ensuring no data ever leaves your computer. Based on Debian Linux, Gnoppix is available with numerous privacy- and anonymity-enabled services free of charge.
What are your thoughts on this? I’d love to hear about your own experiences in the comments below.