What’s next for AI in the social sector?

Generative AI is revolutionizing how we all access information, with a range of positive applications in the social sector. For example, you may have heard us talk about tools we’ve developed like Ask-A-Question, a maternal chatbot providing rapid responses to new mothers’ common questions, or ElectionGPT, which enhanced election efficiency in India by delivering reliable answers to election officials. While these diverse applications helped address very different challenges, one thing they had in common was that they reduced information barriers. 

While such applications have been and will continue to be very useful, they are the lowest hanging fruit and just the beginning. As AI capabilities mature, we’re watching – and actively pushing forward – new areas where AI has the potential to help unlock transformative impact at scale.

From Advice to Action

One of the most promising developments we’re seeing is the ability to move beyond advice and actually help people take concrete actions. With AI agents and “function calling” becoming more robust, we can now help users complete the first steps toward accessing services. Here are a few examples that excite us:

Citizens with Benefits: Many government welfare programs go underutilized because eligible citizens either don’t know about them or find the application process too daunting. Instead of just telling people what benefits they qualify for, AI can now help them actually get signed up by pre-populating forms or directly accessing application portals. Organisations like mRelief in the US and Indus Action in India are already doing this. At IDinsight, we’ve been working with RightWalk Foundation to connect job seekers with an apprenticeship program run by the Central Government’s National Apprenticeship Promotion Scheme (NAPS) in India. Our bot helps people sign up and find matching opportunities simply through a conversation over WhatsApp, a platform they are already comfortable with.

Supporting Farmers: Agricultural advisory chatbots have been around for a while, but new capabilities allow us to go beyond just providing technical information. Now, we can help farmers actually purchase inputs like fertilizer and pesticides, or even sign up for crop insurance—turning advice into immediate action.

Personalization at Scale

The second frontier is about making interventions more effective through personalization. We know that tailored content is significantly more impactful in creating behavior change than generic information. AI makes personalization possible at an unprecedented scale:

Career Advisory: Rather than generic career advice, AI can provide guidance specifically tailored to an individual’s current skills, location, timeline, and aspirations. Organizations like CareerVillage are doing this work in the US, and we’re exploring similar approaches with RightWalk Foundation in India.

CHW Diagnostic AI: While AI question-answering tools for community health workers already exist, tools that can diagnose and recommend treatments for specific patient cases are just emerging. We’ve been working on a Diagnostics Agent that makes expert-level clinic reasoning accessible and actionable at the point of care. 

Personalized Health Advisory: Our partners at REACH Digital Health are taking maternal health messaging to the next level by personalizing content to encourage better health behaviors among mothers and pregnant women.

AI Tutors: Perhaps one of the most exciting applications is personalized education. AI tutors—whether delivered via voice calls, apps, or WhatsApp—can meet students at their current learning level and pace, potentially leading to dramatically improved learning outcomes. Organizations like Youth Impact, Mindspark, ConveGenius and EIDU are all leveraging AI to promote personalized adaptive learning.

Automation for Efficiency

Finally, AI offers tremendous potential for automating or accelerating time-consuming manual tasks, freeing up human capacity for higher-level work, which could unlock impact across multiple sectors. For example:

Education Support: Teachers spend countless hours on lesson planning and materials preparation. AI-assisted lesson creation based on curriculum standards and pedagogical best practices can dramatically reduce this burden while potentially improving quality. Khan Academy’s Khanmigo, and EIDU’s platform offers some teacher support features in this area.

Legal System Efficiency: In India, Adalat AI is tackling the enormous backlog in courts by automating court transcription—a task often performed manually by judges themselves in lower courts. Their solution has reportedly tripled productivity in some settings, allowing judges to focus on adjudication rather than documentation.

Of course, these applications come with their own challenges. As we develop more sophisticated AI tools for social impact, we need to be vigilant about data privacy, equity of access, and ensuring that AI supplements rather than replaces human connections in service delivery.

At IDinsight, we’re committed to approaching these opportunities thoughtfully, focusing on real problems faced by the communities we serve and rigorously evaluating impact. As always, we’re starting with the problems, not the technology—but the expanding capabilities of AI mean we can now tackle problems in ways that simply weren’t possible before.

I’d love to hear your thoughts on these emerging areas or other promising applications of AI in the social sector. What challenges are you working on where these approaches might help? Reach out at sid.ravinutala@idinsight.org

–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––

Sid Ravinutala leads the Data Science and Engineering team at IDinsight, where he oversees the development of AI tools for social impact.