The chapter discusses AI as a transformation agent for inclusive digital education in rural India. While ed-tech has made strides in urban India, India being a developing nation, rural communities are still left isolated with their struggles—lack of infrastructure, low digital literacy, lack of socio-economic opportunities—all leading to accessibility issues in quality education. We propose the RAI (Rural AI for Inclusion) Framework, a contextualised model based on an inclusive pedagogy, participatory approach, and ethical AI implementation. Drawing on mixed-methods research in three districts of Karnataka, the study assesses the deployment of low-cost AI tools implementing surface accessibility in vernacular and support for teachers. Findings show benefits regarding student engagement and instructional efficiency, as well as barriers related to infrastructure and ethics. The chapter ends with policy recommendations for implementing AI at scale, with equity and cultural responsiveness in regions with service access gaps.
Global supply chains are undergoing a paradigmatic shift as climate action, social accountability, and economic resilience emerge as non-negotiable imperatives. This chapter introduces the concept of Triangular Transformation, a comprehensive framework that integrates these three pillars into the design and governance of modern supply chains. Traditional supply chain models, which prioritised efficiency and cost minimisation, have proven inadequate in addressing the interconnected crises of climate change, social inequities, and systemic vulnerabilities. The proposed framework emphasises that achieving sustainability requires simultaneous commitment to decarbonization, ethical labour practices, and resilience-building mechanisms.
The Western Ghats of India, represented by a verified operational boundary and a quality-controlled 0.25° modelling grid. From the 1980–2025 integrated dataset, 100 grid cells satisfied the native completeness criterion and were retained for primary modelling. We evaluate a simulated federated-learning framework for one-day-ahead rainfall prediction using rainfall at day t together with Tmax, Tmin, 850-hPa zonal and meridional wind, 850-hPa relative humidity and sea-level pressure. Five geographic clients train local neural networks and a server aggregates their parameters using FedAvg. Temporal testing uses 2022–2025; spatial generalization is assessed on 20 grid cells withheld completely from training. New Hydrological Insights for the Region: On the 20 unseen cells, FedAvg achieved RMSE 9.633 mm day−1, MAE 4.476 mm day−1 and R2 0.517, compared with 9.953, 4.895 and 0.484 for the centralized neural network and 11.519, 4.855 and 0.309 for persistence. Three-seed temporal robustness gave RMSE 12.562 ± 0.014 mm day−1 and R2 0.613 ± 0.001. Paired bootstrap and permutation tests supported lower aggregate FedAvg errors on the spatial holdout. However, performance varied among cells and FedAvg did not have the lowest P90–P99 errors. The study therefore demonstrates useful spatial transfer under the defined eastern-edge holdout experiment, while recognizing that the federated environment is simulated on one computing system and is not an operational multi-institution forecasting deployment.