<p>Bicycling offers significant environmental and health benefits, making it a sustainable mode of transportation. However, the absence of proper bicycle facilities poses considerable challenges for bicyclists navigating alongside motorists. This issue is particularly pronounced at unsignalized intersections in developing countries with diverse traffic conditions. Assessing these intersections is vital to identify deficiencies and implement effective improvements. Despite this need, suitable models for such contexts remain lacking. To address this gap, this study develops bikeability assessment models using advanced artificial intelligence (AI) techniques. Extensive data from 76 intersection approaches across six Indian cities are analysed to identify the key inputs for the model. The primary variables include bicycle delay, conflicting traffic volumes, commercial density, and more. Efficient techniques such as Functional Network (FN), Genetic Programming (GP), and Multi-Gene Genetic Programming (MGGP) are applied to develop the models. The obtained models demonstrate high prediction accuracy, outperforming traditional methods. Notably, the MGGP model achieved coefficient of determination (<i>R</i>²)-values above 0.92 with the training and testing data. This model rates intersection approaches on a letter scale from A (excellent) to F (worst). Model applications reveal that approximately 86% of intersections currently provide average to poor service levels, emphasizing the urgent need for targeted interventions. Sensitivity analysis identified effective strategies to enhance bikeability, including reducing bicycle delays and increasing the effective approach width. These strategies, combined with the proposed models, provide valuable tools for urban mobility specialists and infrastructure designers to establish safer and more accessible bicycling environments.</p>