A Culturally Responsive and Affect-Aware Conceptual Framework for AI-Assisted Learning in India’s Multilingual Classrooms
This paper proposes SETU, a culturally responsive and affect-aware conceptual framework that bridges India's multilingual classrooms and formal curricula by integrating six components to augment teachers and align AI-assisted learning with the National Education Policy 2020.
Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
In classrooms around the world, learning is often treated as a simple transfer of facts from a teacher to a student. For decades, educational technology has tried to speed up this process by using computers to tailor lessons to individual needs. These systems, known as intelligent tutors, usually work by tracking how fast a student answers questions and what score they get. They then rearrange the order of the material, hoping to keep the student moving forward. However, this approach often misses a crucial part of the human experience: the cultural and linguistic world the student lives in. It assumes that a child learns best in a vacuum, ignoring the fact that students bring their own languages, family traditions, and community knowledge into the classroom. When technology ignores these deep roots, it can accidentally make learning harder for children who speak different languages or come from different backgrounds, rather than helping them.
This gap is particularly wide in India, a country with an immense variety of languages and cultures. Here, many children are taught in a language that is not their mother tongue, creating a structural barrier to understanding. A new conceptual paper proposes a different way forward, not by building a smarter computer, but by building a bridge. The researchers introduce a framework called SETU, a name that means "bridge" in several Indian languages. This is not a finished software product that can be downloaded today, but rather a detailed architectural blueprint for how future artificial intelligence should be designed to respect and utilize a student's culture. The authors argue that for technology to truly help, it must stop treating a student's language and background as background noise and start treating them as the very medium through which learning happens.
The core idea behind SETU is that learning is not just about memorizing facts; it is a social activity deeply tied to who a person is and where they come from. The framework suggests that an AI system should act like a skilled teacher who knows how to connect a child's home life with school lessons. To do this, the proposed system is built around six working parts that talk to each other constantly. First, a linguistic engine recognizes that a student might think in one language but need to learn in another. Instead of forcing the student to switch languages immediately, this part of the system allows them to use their full range of languages to make sense of new ideas, a practice known as translanguaging. It can explain difficult words in the child's home language or accept answers in that language, slowly building a bridge to the language of instruction.
Second, the system includes a mapper that looks for the "funds of knowledge" a student already possesses. This means it identifies the valuable skills and stories a child brings from their family and community, such as farming techniques, local crafts, or oral storytelling traditions. When the system needs to explain a new concept, like a math problem or a scientific principle, it uses these familiar examples instead of abstract ones that might feel foreign. If a child learns best through stories about their village, the system uses those stories to teach. Third, an affect and context sensor watches for signs of how the student is feeling. It looks at cues like frustration, boredom, or fatigue, and considers the environment, such as the time of day or whether the student is learning in a noisy room. If the system senses the student is overwhelmed, it can pause or change the tone of the lesson, just as a human teacher would.
These three sensing parts feed into a central profile that builds a picture of the learner over time. This profile is then used by an orchestrator to plan the next steps of the lesson, ensuring the difficulty is just right and the examples are culturally relevant. A feedback composer then delivers the response, turning corrections into conversations rather than judgments, and speaking in a tone and language that the student finds encouraging. Crucially, this entire process is wrapped in a guardrail for ethics and equity. This component acts as a conscience for the system, constantly checking for bias, ensuring privacy, and making sure the technology does not make unfair decisions. It also ensures that a human teacher remains in the loop, acting as the final judge and cultural mediator, rather than being replaced by the machine.
The researchers built this proposal on real data that highlights the urgency of such a design. They point out that while India has thousands of spoken languages, schools often use only a handful of languages for instruction, leaving the majority of children to learn in a language that is not their own. This mismatch creates a significant disadvantage, particularly for children from tribal or minority communities. Furthermore, recent surveys show that while many rural children now have access to smartphones, they often use them for social media rather than learning, and many still struggle with basic reading skills. The data suggests that simply giving children access to technology is not enough; the technology itself must be designed to meet them where they are, in their own language and cultural context.
The paper does not claim to have solved these problems with a finished product. Instead, it offers a vision for how to build the next generation of educational tools. It suggests that the path forward lies in moving away from systems that simply track scores and toward systems that understand the human being behind the screen. By grounding artificial intelligence in the theories of how people actually learn—through social interaction, cultural context, and emotional connection—the SETU framework proposes a way to make technology a partner in education rather than a distant, impersonal force. The authors emphasize that this approach is not about replacing teachers, but about giving them a powerful tool to extend their reach, allowing them to focus on the deep, human work of connecting with students while the machine handles the routine scaffolding.
Ultimately, this work is a call to designers, policymakers, and educators to rethink the role of culture in learning. It argues that true personalization cannot happen without cultural responsiveness. If artificial intelligence is to be a force for good in diverse classrooms, it must be built to see the learner's culture not as a barrier to be overcome, but as a resource to be celebrated. The SETU framework provides a map for this journey, suggesting that the future of education lies in building bridges between the worlds children live in and the knowledge they are asked to learn.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.