Moumita Saha is nineteen years old, lives in a village outside Kolkata, and each morning before attending college she records sentences in Bengali on her phone. The work is voice data collection for AI training, and the pay is considerably above what most rural work in her area offers. She has described it as the first time she has felt capable of doing something genuinely new. Her experience is not unique. It points to something wider that is happening, somewhat quietly, in the relationship between artificial intelligence and rural women across India.
Most coverage of AI in India focuses on technology companies, urban talent, and large valuations. That picture is accurate as far as it goes. But alongside it is a different story, one that receives less attention and may prove more consequential for the majority of Indian women: the slow, uneven, but real turn of AI tools towards inclusion.
Language as a Barrier, and as a Bridge
Language has been among the most durable barriers that rural women face in using digital services. Government portals, banking apps, and health helplines have historically been designed for users literate in English or standard Hindi. For women in Assam, interior Tamil Nadu, or tribal Chhattisgarh, where neither is a first language, this was not merely inconvenient. It made whole categories of service effectively inaccessible.
BHASHINI, the Government of India’s national AI language platform, is a serious attempt to change this. The platform covers the 22 scheduled Indian languages and a significant number of dialects, and has been integrated into over 23 government programmes. In practical terms, a woman in rural Assam can now ask about her PM Kisan entitlements in spoken Assamese and receive a response in Assamese. She does not need to find someone who reads English or knows how to navigate a government website. The query and answer both happen in her own language.
Suman Sakhi: A Health Companion at Any Hour
In September 2024, the Madhya Pradesh government launched Suman Sakhi, a WhatsApp-based AI health chatbot for women. The name translates roughly as “good friend.” It runs continuously in Hindi and covers pregnancy care, warning signs in high-risk pregnancies, newborn health, and menstrual wellbeing.
Crucially, the developers chose WhatsApp rather than building a separate application. Women in rural Madhya Pradesh were already using WhatsApp; it was already on their phones, already familiar. That decision made adoption far more likely than any standalone app would have. There is a broader lesson here about designing AI tools for communities that are not yet fully digital: entry points matter enormously.
Women as Producers, Not Only Recipients
One aspect of this story that tends to be overlooked is that rural Indian women are not only the intended users of AI systems. Many are also the people whose labour makes those systems possible.
Karya, a Bengaluru-based social enterprise, employs rural women to produce the training data that underlies AI language models. This involves recording voice samples, translating phrases, and annotating audio in regional languages, all paid above prevailing market rates. Karya now employs over 30,000 women across six languages. Moumita Saha is among them. The Bengali data she produces will eventually be used to train the very AI systems that communities like hers will rely on.
A 2024 study of Farmer.Chat, an AI agricultural advisory service running across twelve Indian states, found women returning to the platform more frequently than men and engaging more deeply with its content. Among women users in India, 61 per cent reported improved quality of life within 45 days. More recent 2025–26 data puts women at 35 to 40 per cent of users, with engagement two to three times that of male users.
These are women who have long had less access than men to expert agricultural advice. The gap Farmer.Chat is closing is not technical. It is a gap in who was considered a relevant user in the first place.
Self-help Groups and the Lakhpati Didi Pathway
In Self Help Group meetings across Jharkhand, Odisha, and other states, members are increasingly using AI-assisted tools in livelihood decisions: checking real-time mandi prices before agreeing to sell, consulting agri-advisory tools in their own language on crop planning, and using digital payment records to build a credit history.
The government’s Lakhpati Didi initiative, under DAY-NRLM, targets sustained annual household income of one lakh rupees or more for SHG members. As of February 2026, 3.01 crore women have reached this threshold, meeting the previous target ahead of schedule. A revised target of six crore by 2029–30 has been set, backed by a 20 per cent increase in DAY-NRLM allocation in the 2026–27 Union Budget.
None of the tools involved are technically sophisticated. Price alert applications, WhatsApp crop advisories, and basic digital payment systems are modest instruments. What they address is a long-standing and quite fundamental problem: women selling produce have historically had no independent way to know what their goods were actually worth.
A middleman who knows the mandi rate and deals with a seller who does not has a structural advantage that has little to do with market efficiency and a great deal to do with information access. These tools shift that balance, modestly but meaningfully.
What Remains to Be Done
None of this warrants uncritical enthusiasm. The digital divide is real and affects women disproportionately. Data costs remain prohibitive for many rural households. Digital literacy is uneven by age, region, and caste.
Several of the programmes described above are still in early deployment and have not been tested at full national scale. AI systems trained on limited or unrepresentative data will reproduce existing biases unless developers actively work against that tendency, and evidence suggests this does not happen automatically.
India is building AI infrastructure with multilingual design as a core requirement, not an afterthought. Platforms are being deployed through channels women already use. Programmes are being evaluated, at least in part, by whether they reach women who were previously unreachable.
Whether this direction holds, and accelerates, is a question that researchers, policymakers, and the communities involved will need to keep asking.
Sumber online:
