If the 19th century witnessed armies invading nations across the continents, colonising lands and people with the help of guns and ammunition, in current times, we are experiencing colonisation of another kind—through digital means, trying to enforce dominance in the virtual domain.
Virtual domination is attained through large-scale deployment of cloud infrastructure, proliferation of social media platforms, easy-to-download apps and search engines that have the capability of influencing the habits of information access.
Today, AI algorithms and AI models are shaping national discourses through cultural and political interventions via elections and social media posts, influencing the thoughts and voices of citizens. Continued digital domination would lead to profits reaped from local activities being amassed abroad.
Cause for concern for the global south
Further, generative AI models are trained primarily using Western data and therefore, they are not designed to relate to Indian languages and cultural nuances. As a result, biases crop up, and the representation of the diverse people of India is not reflected in these models.
Some ASEAN and African countries use AI tools that store sensitive patient data outside the country, leading to compromise of patient privacy and the inability to build health systems addressing the localised diseases.
Several universities in Latin America use foreign AI education platforms and content provided along with them, thus creating intellectual dependency. Concerns for protecting the nation’s sovereignty have prompted several countries to promulgate laws to create protection for personal data and also initiate steps to promote local digital infrastructure and digital solutions.
Open source LLMs, SLMs and open AI models are being considered as alternatives to US-led models by several countries. Examples include the Mistral initiative of France, the UAE government initiative-Falcon LLM and the Indonesian initiative Nusantara AI Project.
What should be India’s next steps for fostering AI sovereignty
In India, steps initiated for fostering AI sovereignty include the framing of the Data Protection Act and developing India stack, fuelled by Aaadhar identification, ONDC, Digilocker and also the Sarvam AI initiative, to name a few. Bhashini and Bharat GPT programs are good foundations for supporting the multilingual requirements.
India would have complete AI sovereignty when its digital architecture enables it to build its AI models and all computations are carried out on its infrastructure. Sovereignty does not mean isolation. Careful balancing of independence and conscious collaboration should be the cornerstone of Indian strategy.
