Source: India Today

Amidst the AI fanfare, it becomes extremely important for budding policymakers, scholars and researchers to debunk the myths surrounding AI and to treat it as a technological tool capable of effectively performing some chores, while being equally likely in creating errors. The article examines the Union Budget to reflect on the status of AI’s progress in India, alongside the Impact AI Summit.

The Rise of AI and India’s Policy Agenda

Until now, 2026 has proved to be a pivotal year in India’s AI journey. The Budget Speech made by the Hon’ble FM in February invoked the use of the term “AI” eleven times, the highest till date, an indication of its growing importance. Later that month, India hosted the five-day Global AI Summit which brought forth serious discussions about AI, policy and society to the general public. Both developments indicate the country’s enthusiasm for the development and deployment of AI technologies across sectors. Artificial Intelligence technologies, however, are fraught with grave faultlines such as biases within the technology, hallucinations, environmental and safety concerns; governing the same, therefore, poses serious policy questions that governments across the globe are currently grappling with.¹

1 India Budget | Ministry of Finance | Government of India. (n.d.). https://www.indiabudget.gov.in/

Union Budget vis-à-vis AI:

The Union Budget 2026-27 envisions the integration of AI across various domains, from agriculture and education, to the scaling up of production of assistive devices for persons with disabilities. The India AI Mission, as per the Budget Estimates for the year, is allocated Rs.1000 crores for the FY, which is a direct 50% cut in light of the Ministry’s limited absorption capacity and past trends. As per a response to a question on Lok Sabha in December 2024, over 85% of the ₹1,100 crores in subsidies disbursed under the compute pillar of the IndiaAI mission has been directed towards the development of indigenous generative AI models like Sarvam, BharatGen, Zenteq, SoketAI, Gnani.ai, and Avataar AI, which have been the recipients of these subsidies- all of which will result in increased demand for domestic data centres.²

When asked to generate the image of an Indian in May 2024, MetaAI displayed a peculiar predisposition to generating a ‘man with a turban’ – almost four out of five times.³ This depicts an inherent bias in the model, despite India’s rich cultural and demographic diversity, a fact ostensibly attributable to the biases present in the training dataset. Whether indigenous models could counter the bias that Western models display is yet to be determined. Pratyush Kumar, the co-founder of Sarvam, called for the development of such indigenous technologies to protect our citizens from data colonialism.

Incepted at Bletchley Park in the UK in 2023, the first Impact AI Summit began with a clear agenda of understanding the risks that frontier AI models pose, and designing strategies to counter them. With AI developing as quickly as the models themselves, the governance of AI becomes all the more vital. The Indian PM, alongside other world leaders, thus called for a more inclusive, trusted and collaborative AI, while positioning India as the tech leader of the Global South.⁴ The Summit poses larger questions about the deployment of AI revolving around safety regulations, the impact on marginalized communities, labour, and attempts to ‘govern’ it.⁵

2 Decoding the budget for India’s AI priorities – Takshashila Institution. (n.d.-b). https://takshashila.org.in/content/blogs/20260203_ai-mission-budget-analysis.html

3 Mehta, I. (2024, May 8). Meta AI is obsessed with turbans when generating images of Indian men. TechCrunch. https://techcrunch.com/2024/05/07/meta-ai-is-obsessed-with-turbans-when-generating-images-of-indian-men/

4 India AI Impact Summit 2026. (n.d.). India AI Impact Summit 2026. https://impact.indiaai.gov.in/

5 Kahekashan, & India, H. (2026, February 18). India pushes for sovereign AI at global AI summit. The Hans India. https://www.thehansindia.com/tech/india-pushes-for-sovereign-ai-at-global-ai-summit-1049501#

The Limits of Technological Solutions

In the Budget presented by the Hon’ble Finance Minister, the government announced a High Powered Standing Committee meant to look at possible pathways from education to employment and entrepreneurship, which will identify services sub-sectors with growth potential, assess the impact of emerging technologies including AI on jobs and skills requirements, while proposing “specific measures” for including AI in the education curricula from school onwards and upgrading State Councils for Educational Research and Training Institutes for teacher training.⁶ While many AI-skilling programs are run by the government and private tech companies, their impact needs to be critically analysed keeping in mind the nature of the use to which AI will be deployed, or even whether AI as a technology is ready to be rolled out to young adolescents who might unintentionally inherit its biases, jeopardizing their safety as well. Further, in the Indian context, marginalized communities are often excluded from such programs due to inadequate infrastructure, accessibility and affordability constraints. Thus, will AI bridge the digital divide prevalent in India or widen the same remains to be seen.

The implicit concerns underlying these models can be understood through an example. Described as a multilingual AI tool that shall integrate the AgriStack portals and the ICAR package on agricultural practices with AI systems, the Bharat-VISTAAR model will enhance farm productivity, enable better decisions for farmers and reduce risk by providing customised advisory support.

While the initiative promises to empower farmers with real-time, multilingual information on insurance procedures, weather risks, and claim status, it ignores the array of concerns that shadow the Indian agricultural space today. From divided digital literacy among farmers to challenges such as fragmented landholdings, the exclusion of landless farmers from databases, inconsistent data sources, unclear land ownership and legal disputes, the initiative fails to recognize the grassroot reality of Indian farmers, who are a diverse community.⁷ The solution to the more pervasive concerns among farmers may not lie in technology-based solutions, but rather in questions of adequate minimum support prices and subsidies. Data heterogeneity, which is a defining feature of the diversity of Indian agriculture, does not speak in standard codes. Pearl millet is bajra in Hindi, bajri in Gujarati, kambu in Tamil, and sajje in Kannada.

6 Lakshman, A. (2026, February 1). Education and skilling get Budget push; high-power panel to look at education to employment pathways. The Hindu.

https://www.thehindu.com/business/budget/education-and-skilling-get-budget-push-high-power-panel-to-look-at-e ducation-to-employment-pathways/article70578173.ece

7 Garg, R. (2022, February 22). #StandWithFarmers: Joint Letter by 55 groups | Hold the Agriculture Ministry and Microsoft India accountable. Internet Freedom Foundation (IFF).

https://internetfreedom.in/joint-letter-to-the-agriculture-minister

Such heterogeneity could be misinterpreted by AI models or they may even fail to generalise across regions if such linguistic and regional nuances are not considered.⁸

The Way Ahead:

India needs to develop technologies tailored to its own socio-economic needs which solve the problems of the nation, rather than deploying AI technology wherein deeper structural issues exist (such as agriculture).⁹ This creates a risk of misallocation of public resources. A better approach moving forward is to examine the actual need for the deployment of AI technologies, rather than taking a ‘one-size-fits-all’ approach.¹⁰ India, therefore needs a people-centred AI policy that exhibits the spirit of democratic and participatory governance. With more emphasis on locally-built AI models like Sarvam-AI, India can compete on the global stage while safeguarding its users’ data and interests. Models like Sarvam can be designed keeping in mind a techno-legal framework wherein legal frameworks and safeguards are embedded within the technology itself. India has the potential to lead global AI development by prioritizing purposeful AI, built around societal welfare, rather than competing for technological power. Existing initiatives such as E-Mitra chatbot for farmers or the eSanjeevani telemedicine platform are major breakthroughs in the agriculture and healthcare sectors respectively, and are useful guides for AI developed for aid in specific fields.¹¹ Deployment of Small Language Models (SLM) based for specific purposes can be cost effective, efficient and flexible as per that particular domain. While doing so, the country’s environmental and resource constraints should be constantly evaluated so as to not put the strain of technological advancement on the environment. In a resource-constrained and climatically vulnerable context as that of India, it is of crucial importance to regulate the expansion through adequate safeguards, efficiency standards, and sustainability assessments.¹²

8 Contributors, E. (2026, March 6). Artificial intelligence in Indian agriculture: The bus India cannot afford to miss.

The Economic Times.

https://economictimes.indiatimes.com/news/economy/agriculture/artificial-intelligence-in-indian-agriculture-the-bu s-india-cannot-afford-to-miss/articleshow/129168914.cms?from=mdr

9 Frontline News Desk. (2026, February 23). India AI Summit 2026: Who controls future tech order? Frontline. https://frontline.thehindu.com/science-and-technology/india-ai-summit-pax-silica-tech-geopolitics/article70660408

.ece

10 Vipra, J. (2024, October 10). What would a people-centred AI policy for India look like? Economic and Political Weekly. https://www.epw.in/journal/2024/40/insight/what-would-people-centred-ai-policy-india-look.html

11 Contributors, E. (2026, March 6). Artificial intelligence in Indian agriculture: The bus India cannot afford to miss. The Economic Times.

https://economictimes.indiatimes.com/news/economy/agriculture/artificial-intelligence-in-indian-agriculture-the-bu s-india-cannot-afford-to-miss/articleshow/129168914.cms?from=mdr

12 Renu Khandelwal, “Small Language Model: LLM that Does More with Less,” Medium (12 August 2025), available at: https://arshren.medium.com/small-language-model-llm-that-does-more-with-less-b708573f3a7f

Key takeaways:

• India’s AI policy must prioritise people, keep in mind the unique context the country is placed in and develop technologies that align with local realities.
• Sustainable AI leadership for India would mean balancing between technological ambition and its socio-economic and ecological constraints.
• Therefore, a snake-oil approach towards AI may lead to misallocated resources and overlooking deeper socio-economic issues.

Conclusion:

AI technologies are dynamic and so are the challenges that come embroiled with it. Thus, as India posits itself as a leader of the Global South in AI technologies, it should do so keeping in mind that the technology facilitates in alleviating the problems of the people on ground. This would require a shift in the policy orientation- one from promoting AI as an end in itself to the one that helps in achieving the broader goal of public welfare, well-being of the citizenry and sustainable development.

Other references:

1. Das, K. (2026, February 1). Budget 2026: Big infra, AI data centres, defence push, cheaper cancer drugs.                                                                India                                                                          Today. https://www.indiatoday.in/business/budget/story/union-budget-2026-infra-defence-ai-cancer-drugs

-tax-push-key-details-2861247-2026-02-01

• AP, B. (2026, March 5). From data to Decisions: What Bharat-VISTAAR needs to transform Indian                           agriculture                 –                 The                 Wire.                 The                           Wire. https://thewire.in/agriculture/from-data-to-decisions-what-bharat-vistaar-needs-to-transform-india n-agriculture
• Deep, A. (2026, February 23). What are the key takeaways from the AI summit? | Explained. The Hindu.

https://www.thehindu.com/sci-tech/technology/what-are-the-key-takeaways-from-ai-summit-expl ained/article70661026.ece

Jindal Policy Research Lab ( JPRL)

The Policy Research Lab provides non-partisan facts that will create informed public opinion and enhance appropriate decision making for citizens at large. The lab’s principal objective is to provide policymakers and organisations with empirically supported guidance on policies that will improve people’s lives.

The Policy Research Lab aims to provide robust data and analyses on various aspects of policy making, through survey research, interviews, focus-group discussions, case studies and any other methodology conducive and appropriate for lab settings. The lab adopts a non-advocacy approach, and its data are factual and as they exist.