<p>The agriculture sector is becoming susceptible to climate change in Pakistan, where the shrinking crop yields, increased input prices, and poor digital connectivity are notable barriers to sustainable production. This paper examines the role of Information and Communication Technology (ICT) in grain yields and climate-resistant rice production in Pakistan between 2002 and 2022. It uses a hybrid approach in terms of AutoRegressive Distributed Lag (ARDL) modelling and machine learning (ML) techniques (Linear Regression, Ridge Regression and Random Forest) in order to gain both causation and prediction. The ARDL findings indicate that internet use has a statistically significant long- and short-term positive impact on total grain output, though mobile subscriptions are only significant in the short term. Other ML models verify important socio-economic and climatic variables, including fertilizer utilization, ease of accessing credit, irrigation facilities, and water pressure, to discover that water supply and fertilizer use are significant determinants of rice productivity. Interestingly, Random Forest and Gradient Boosting are better applied to predicting agricultural outcomes in complex environments than conventional regression models. By combining macro-level ICT indices with micro-level resilience elements and providing a data-based approach to digital agricultural development, this two-step methodology fills in the literature gaps related to earlier works on digital agricultural development. The results highlight the urgency of scaling digital advising services, precision irrigation, and climate-smart tools to support food security and sustainability. The study is also useful to policymakers interested in enhancing agricultural adaptive capacity to climate risk and technology limitations in Pakistan.</p>