AI summit: Broad consensus, soft impact

Beyond the regular fanfare that accompanies tech CEOs and heads of state when they descend on a city, the India AI Impact Summit saw wide participation across stakeholders, announcements of big investments, and a Summit declaration signed
by 91 governments, including the United States, the United Kingdom, and China. While these are numerically more signatures than the previous three summits, the text is weaker as
it hinges on ‘voluntary and non-binding’ commitments.

This was the fourth in a series of global AI summits; the previous editions were held in Bletchley, Seoul, and Paris. These summits are forums for international coordination on AI governance and play a crucial role in deciding the future agenda.

The India AI Impact Summit, the first in the Global South, was a pivotal moment for spearheading AI governance discussions and setting an agenda that represented the needs of the region. In a world where a few global corporations, mostly in the Global North, control significant resources for building AI technologies, the Indian government took critical strides in placing democratic access and AI sovereignty on the table.

In line with previous summits, the main declaration rightfully recognises the importance of international cooperation and multistakeholder engagement in global AI governance. The declaration is framed around seven themes identified by the Impact Summit: democratising AI resources, economic growth and social good, secure and trusted AI, science, access for social empowerment, human capital, and resilience, innovation, and efficiency. Across these themes, there is significant emphasis on AI diffusion (spreading use of AI), ‘voluntary and non-binding’ frameworks, and international collaboration.

While the Summit witnessed inclusive participation, the declaration was decided and shaped by nation-states with limited engagement and input from civil society and academia. Building towards a wide consensus across States in this fraught geopolitical moment has clearly resulted in a weakened document that relies on voluntary commitments that don’t meaningfully safeguard the most vulnerable groups in the Global South. Reflecting on the outcomes of these summits provides a view into changing priorities on global AI governance.

Users in the Global South will feel higher and more significant impacts of AI (both positive and negative) and it is crucial that we build a human-centric approach that enables user safety. Last year, India co-chaired the Paris AI Summit with France which emphasised a human rights-based, human-centric, ethical, safe, secure and trustworthy deployment of AI technologies in its final statement. In the New Delhi declaration, despite focusing on ‘benefits to humanity’, there is no emphasis on human-centric fundamental freedoms, or rights-based approaches to AI, a language found in all previous Summit declarations. Even the phrase ‘human-centric’ appears only in the Prime Minister’s opening address and not the declaration.

To safeguard the interests of the Global South, India should have carried forward this language from the previous declarations into the final consensus-driven document.

Leading the Global South

India has consistently championed linguistic diversity as a primary goal of its AI governance strategy to enable the inclusion of a diverse set of users and bridge the digital divide. Without Indian languages forming part of the input data for popular AI models, they will remain inaccessible to a large section of society. At the Impact Summit, India released a voluntary commitment, agreed upon by domestic and global AI companies, on strengthening multilingual evaluations for the availability of AI systems across languages, cultural contexts, and use cases. But curiously, unlike the Paris and Seoul AI summits, the New Delhi declaration contains no reference to linguistic diversity. This seems to be a missed opportunity for India to demonstrate its leadership as a multilingual Global South nation.

Right before the India Impact Summit, India amended rules under the IT Act relating to synthetically generated content (deepfakes) that harm women and children. But again, unlike Paris and Seoul, the New Delhi declaration does not lay down guardrails on gender or child safety, diversity, and equality. This is puzzling as the moment to acknowledge AI-related harms specifically on children was ripe given the global momentum around child safety in various regions including Australia, Brazil, China, the European Union, France, Germany, the UK, and the US.

This is particularly glaring, given India had an expert engagement group focused on child safety in AI in the runup to the Summit.

The lasting image from the India AI Impact Summit is that of tech CEOs with the Indian Prime Minister in a show of strength. While it’s commendable for India to showcase its partnership with industry, a more symbolic image of India’s push for AI sovereignty and democratising access to AI would have been heads of state from the Global South coming together in solidarity.

South-South alliances like that with Brazil help centre the needs of our communities over geopolitical rhetoric. Later this year, India has an opportunity to spotlight the needs of users from the Global South in key global governance forums such as
the UN AI Global Dialogue and the BRICS Summit, and demonstrate meaningful Global South leadership. These present crucial opportunities for India to push for bolder demands on multilingual access, equitable access to rare earth minerals, gender and child safety and diversity, and charting a ‘third-way’ for AI governance. Here, India should demonstrate true Global South leadership.

(Shashank is the Associate Director and Jhalak is the Executive Director of the Centre for Communication
Governance at National Law
University Delhi)

Disclaimer: The views expressed above are the author's own. They do not necessarily reflect the views of DH.