India seeks consensus in GPAI Summit talks

At a pre-summit press briefing, Vaishnaw said the declaration aims to advance talks on global AI regulatory frameworks, besides pitching for a digital public infrastructure (DPI)-like model that could be adopted by participating nations.

“There will be regulatory aspects that are in line with past agreements and declarations. The thinking process of GPAI will be in line with global ideas. We are negotiating a declaration document at the end of GPAI 2023, which we hope we’ll be able to arrive at through consensus. The global thinking on AI is converging, and people understand the dangers, potential and the benefits of AI, but the world also wants to put certain guardrails around the development of AI. We will be aiming at a statement that is agreed to by everyone.”

The minister said the GPAI declaration will have two aspects highlighting India’s stance around AI. “The first is to evaluate the use of AI in sustainable agriculture, adding to the previous GPAI themes including healthcare, climate action and building a resilient society. The second is on collaborative AI—in line with our DPI approach.”

India’s DPI model forms a key basis for the Delhi Declaration adopted by all members of the G20 Summit in September and was offered as a framework that could be replicated globally. Experts described the DPI model as India’s “soft power” strategy among major global economies. “In DPIs, there is a common platform built using public funds. Banks, startups, markets and users joined this platform, which binds all stakeholders. In such a public-private partnership, everyone can access technologies. A similar approach will be taken towards AI, in order to create a public platform where entities such as MSMEs, agriculture universities, climate-tech startups and so on can access.”

Vaishnaw said this approach will look to reduce the overall cost of technology platforms, thereby promoting “democratization” of technology access.

India was one of the 28 signatories to the Bletchley Declaration—a global policy approach agreed upon by the US, the UK and China, along with India, at the UK AI Safety Summit last month. Union minister of state for IT, Rajeev Chandrasekhar, represented India on AI at the summit.

Explaining India’s additions to the declaration following the UK Summit, Chandrasekhar said: “We spoke about the need to have safe, trusted AI platforms, and distinguish them from unsafe, untrusted platforms. We proposed that AI should not be demonized in India or around the world, as it represents a massive opportunity. We also spoke about who will determine safety and trust, and discussed four harms that come out of AI—workforce disruption, privacy impact on individuals, harms that are non-criminal, besides weaponization or criminalization of AI.”

“The Summit will largely discuss points that are in line with global issues around the development of AI,” Vaishnaw said adding that the India AI platform developed by the ministry of electronics and IT in October, will not be directly involved in issues set to be discussed at GPAI 2023.

The Indian AI platform approach seeks to develop an ‘India datasets platform’ that will build and maintain an organized database for local Indian languages, to be accessed by industry stakeholders, researchers and academia. The second part will involve developing indigenous compute infrastructure to power development of AI models, tuned specifically for India’s use cases, in a public-private partnership model. Industry stakeholders have said that developing India-specific AI models will be crucial in order to address issues of bias that are specific to India—which global models such as OpenAI’s Generative Pre-trained Transformer (GPT), Meta’s Llama or Google’s Gemini may not directly cater to.

India’s regulatory approach to AI has so far been centred around AI harm. On 7 December, S Krishnan, secretary at Meity, said at a press conference that India will take a “light touched regulatory approach” towards imposing guardrails around AI. The upcoming GPAI Summit, Krishnan added, will further seek to represent “more of the global South” in global AI regulatory development.