Global South Must Build Its Own Multilingual AI Models, Says Amitabh Kant at IndiaAI Summit

Concerns over data colonialism and unequal access to artificial intelligence took centre stage at the IndiaAI Summit, with policymakers from India and Africa calling for a more inclusive global AI roadmap.

Speaking at the summit, Amitabh Kant, former CEO, Niti Ayog, said India alone is contributing nearly 33 percent of the data that global companies use to refine large language models. These models, he noted, are being improved using data from the Global South and could eventually be sold back at high prices.

He argued that if countries in the Global South are powering the refinement of AI systems, they must also build models that are trained on their own data and serve their own developmental needs.

Kant described AI as a transformational technology that will disrupt every sector and way of life. However, he cautioned that without deliberate policy choices, it could deepen inequality. AI must be accessible, accountable and multilingual, he said, warning that systems built only for a few dominant languages would exclude large populations.

The real test, he added, is whether AI can meaningfully reach people at the bottom of the pyramid and improve education, livelihoods and public services.

Drawing parallels with India’s digital public infrastructure, Kant said the country’s success in fast, low-cost digital transactions was driven by open APIs and open-source frameworks that enabled private innovation. A similar “phygital” public AI layer is needed to ensure that technology creates a more equal society rather than concentrating wealth and market power.

He also pushed back against the idea that regulation and innovation are at odds. The real divide, he said, is between those who have access to AI opportunities and those who do not. Smart guardrails can enable innovation while ensuring accountability, especially in sensitive sectors such as financial markets that require human oversight.

Echoing these views, Amandeep Singh Gill, Special Envoy for Technologies, United Nations said the summit marked the first major coming together of Global South nations on AI, stressing that inclusive coordination is essential to prevent accountability gaps. He pointed to ongoing discussions at the United Nations around a three-pillared approach grounded in the science of AI.

Arunabha Ghosh, CEO of CEEW said while AI is undeniably disruptive, the regulatory guardrails must be inclusive and aligned with broader goals such as decarbonisation and digital transformation.

Leaders from Africa also underlined the need for collaboration. Philip Thigo, Special Envoy on Technology in the Office of the President of Kenya, said access and affordability remain key challenges.

Kenya, he noted, is among the largest users of ChatGPT subscriptions at around $20 per month, which remains expensive for many. He called for differentiated pricing models that reflect the realities of the Global South.

The broader sentiment across the panel was:as AI grows in power and capacity, countries of the Global South must act collectively to ensure the technology narrows, rather than widens, the gap between the haves and have-nots.