EVERY DECADE HAS WITNESSED technological hype, catalysing innovation and automation by shifting systems from manual to digital. Each transition is judged by how far citizens and communities, especially those facing multiple social and digital divides, are left behind in the race towards socio-economic progress. Today, artificial intelligence (AI) is presented as a disruptive technology across every sector, institution, and design, promising to reach where systems have struggled to reach. Yet the reality for rural, underserved, and unconnected field workers and communities points to several failures in technology adoption, compounded by marginalisation that denies them access to the digital world. This piece asks whether AI-enabled nutrition and public health systems are designed, developed, deployed and adopted by factoring women’s interests and perspectives , especially those at the bottom of the social hierarchy and adversely affected by the gender digital divide, as the starting point or as an afterthought.
Women are not a Monolithic Community
The World Health Organisation’s A Global Health Strategy for 2025-2028 report indicates that malnutrition among women and children causes 4 million deaths annually globally. Official statistics on the nutritional status of women in India are evident, showing a high prevalence of anaemia among women, adolescent girls, pregnant women, and children, with numbers exceeding those of previous survey rounds. NFHS-5 data show that 57% of women of reproductive age (15-49 years) are anaemic, while 67.1% of children are also affected by anaemia. Anaemia is more prevalent among women (61.1%) in the Aspirational Districts. The highest anaemia rates are recorded among women from SC/ST, Muslim, rural, and economically disadvantaged households. These data identify the same women who also face the gender digital divide. For example, 31% of women in India have ever used the internet, with 24.6% in rural and 51.1% in urban areas. Among the poorest population, who fall in the bottom 20% of the socio-economic hierarchy, only 9% of women have ever used the internet. Regarding mobile ownership, only 31% of women own a mobile phone, and India has one of the widest gender gaps in the Asia-Pacific region. The lack of literacy, along with barriers related to gender, caste, religion, and demographics, impacts women differently. This evidence shows that the same women facing the highest anaemia burdens also face the lowest digital access. Consequently, women lack access to health and nutrition information, as well as the ability to address digital, social, and health-related harms or to navigate their circumstances with autonomy.
Against this background, while the latest NHFS-6 fact sheet has introduced digital literacy, it has not provided anaemia statistics among many other important indicators, while making the headline: “NFHS-6 reflects India’s accelerated progress in maternal and child health, nutritional and financial protection.” Although the Government cites this as a conscious decision citing the blood-testing method issue, this absence of anaemia estimates from the NFHS-6 fact sheet limits the continuity in nutrition-tracking and the implications for evidence-based planning. Without the crucial data, how can policy-design develop hyperlocal strategies and interventions to address contextual health and nutrition-related challenges? Moreover, as monitoring nutrition and prevention of anaemia have been a huge part of frontline health delivery, what implications will the absence of data have for the scope of work of frontline community health workers?
Frontline Health Institutions in the Technocratic Era
Let us examine the role of Government-designed AAA platform for health and nutrition
service delivery: - the Accredited Social Health Activist (ASHA), the Anganwadi Worker (AWW), and the Auxiliary Nurse Midwife (ANM). These women are frontline community health workers, including 1 million ASHA workers, 1.3 million AWW workers, and 0.2 million ANM workers, all of whom serve as primary access points for communities and therefore form the backbone of health and nutrition delivery in the contemporary technocratic India. Their mandates are huge and diverse, ranging from facilitating maternal and child health services, immunisation support, nutrition counselling, supplementation follow-up, take-home ration distribution, community mobilisation for health awareness, access to information about health schemes, monitoring growth, and data collection and reporting using technology interfaces. It is widely reported that rural women hired as volunteers in the AAA technology-integrated frontline delivery system are overburdened, undertrained, and underpaid (below the minimum wage), while expected to perform care, data, digital, and reporting work. Women from marginalised castes are often subjected to untouchability, exclusion, and prejudice, which undermine their role as frontline community health workers. Amid a lack of safeguards, redressal mechanisms, capacity-building and institutional support, this workforce is treated with disrespect and is often blamed for failures in the digital health system with no appeal mechanism available to them.
Do Data Systems reflect the Ground Realities?
Built on existing institutions and systems and confronted with several basic challenges - gender digital divide, unreliable connectivity, shared mobile ownership, lack of hyperlocal need assessment resulting in no adequate targeted literacy and capacity building, and overburdened field workers - technologies are integrated as add-on layers that link data and identification systems to enable targeted service delivery. Because these data systems are only as good as the data fed into them, their unavailability shapes the inaccessible outcomes. In most of the Asia-Pacific region, data quality is low, incomplete, incomprehensive, and it is firmly entrenched among the more privileged on the socio-economic ladder. For example, the lack of statistical data on anaemia in the NFHS-6 fact sheet breaks the continuity of data essential for evaluating and designing nutrition policies. This evidence helps explain why women, particularly experiencing multiple forms of marginalisation, are underrepresented in data systems and why the resulting technology solutions are unrelatable and exclusive. It also explains why women are not properly trained to keep pace with rapid changes in frontline service delivery, why they lack adequate digital literacy, and why they are overburdened with deliverables that lack sufficient support or training to use technology interfaces.
A major roadblock to social, digital, and health inclusion is also the digital-only policy approach, which entails an unprecedented and often unquestionable and unchallengeable integration of data-intensive technologies into every domain, while undermining the critical role that manual, human-assisted design has played. Notwithstanding the significant gaps in design thinking, contextualisation, hyper-local assessment, representation in datasets, infrastructure, and capacities, most governance models are de facto ready to integrate AI to deliver seamless services. The grassroots realities, however, highlight the status quo, with people struggling to avail entitlements even before the AI hype took over. For example, many communities affected by digital exclusion were not able to use Aarogya Setu during COVID-19. Hence, data on COVID-19 disease spread remained incomplete and underrepresented, leaving out those who needed public health support the most. In its latest move, the POSHAN Tracker app has integrated facial recognition software (FRS), developed by Google, to track the distribution of take-home rations among communities This necessitates that Anganwadi workers use facial recognition and e-KYC, which can create gaps and exclusions, including authentication failures and mismatches due to connectivity issues and inadequate training to navigate this scheme, and data-related concerns that reinforce social biases, leading to the denial of ration.
In rural Jharkhand, a lactating woman and her child were denied supplementary nutrition because FRS did not validate the mother's photograph in the Aadhaar Card. Moreover, as the phone linked to her Aadhaar Card was with her husband, who lived in Delhi for construction work, she could not receive the OTP. Owing to biological changes that pregnant women experience, there are multiple cases where they have not been recognised and validated by the FRS, leading to denial of rations. These realities show a glimpse of why the technocratic, digital-only policy approach could still be missing the picture emerging from on-the-ground evidence.
Just AI for Last-Mile Nutrition Delivery
Digital Empowerment Foundation's (DEF) Just AI - Data & Algorithms for Communities initiative approaches AI through a social justice lens. The fundamental question of this inquiry is: Can we build Just AI? How do we ensure we account for women and resource-constrained contexts in digital design and nutrition AI systems? What should be the criteria to decide whether digital intervention is required at all? DEF’s work with last-mile communities over three decades has given us many lessons. Our proposition is simple and direct. Let us go to the communities first, not last. Public health and nutrition technologies cannot be supply-driven; they must be demand-driven, grounded in an ecosystem approach, bottom-up, women-led, phygital, community-centric design. With a pan-India network of 2400+ Communication Information Resource Centres (CIRCs) led by SoochnaPreneurs (Soochna = “Information”; Preneurs = “Entrepreneurs”) across 28 states, this design thinking has centered women and PwDs with different marginalities, showcasing the role of community-owned infrastructure in strengthening rural digital entrepreneurship and ensuring meaningful access to information, digital services, health services, education, financial services, digital market linkages and livelihood opportunities. Built on six grassroots models - SoochnaPreneur, SmartPur, Udyamini, Internet Roshni, Wireless for Communities, and Samriddh Gram - DEF’s Just AI framework entails a long-standing, community-owned, self-sustainable, environmentally just, and trustworthy social infrastructure. These hyperlocal digital ecosystems, when integrated with AI, have the potential to span multiple layers, ranging from addressing connectivity challenges to ensuring phygital access to digital services, social welfare for target communities, participatory data practices, capacity building, meaningful inclusion, and livelihoods. .
We must question the data. The question we should ask is not what AI can do, but who decides and designs it and who it is designed for, centring the most marginalised are at its heart. AI must be leveraged to turn authenticated data into wisdom for the benefit of people, not just for the intermediary tech companies that house the data. Backed by reliable, trusted, disaggregated data - the nutrition programs, policies, and systems should be designed to serve the poorest women - who are severely deprived of resources that affect their access to sufficient, timely, and nutritious food and a healthy environment. In schools and colleges across the country, nutrition should be studied as part of the curriculum, so children grow up aware of its importance in living a healthy, dignified life. Most importantly, a phygital AI ecosystem built on well-researched, disaggregated, hyperlocal, community-owned, trustworthy, contextualised data must be co-created with 2,50,000+ Gram Panchayats across the country. This ecosystem will thrive only when it has built-in needs assessment, human-assisted design, independent audits, evaluations, accountability checks and grievance redressal mechanisms; an authenticated, trusted call centre with a person on the other end, not a bot; and mobile-friendly access for resource-limited settings.
