In the field of road infrastructure assessment, the emergence of cutting-edge methods such as convolutional neural networks (CNNs) and Computer Vision techniques represents a profound diversion from traditional visual inspection methods. This research compares automated road damage assessments based on these emerging digital methodologies with conventional on-site inspections, specifically for the case of concrete roads. The main objective is to outline the differences in efficiency, accuracy, and cost-effectiveness between these two approaches. Based on a robust data set, covering a variety of pavement sections inspected by professional inspectors, this study contrasts the results of on-site visual assessments with the results obtained from an automated CNN and Computer Vision based methodology. The automated system identifies different types of damage, locates them in the images, and segments the damage to accurately measure its extent. This research explores the advantages and potential limitations of automated systems, focusing on their consistency, objectivity, and implications in terms of resource and infrastructure management. Emphasis is placed on the variability inherent in manual inspection processes, uncovering discrepancies and potential biases that can hold back the accuracy and reliability of traditional methods. This research responds to the growing need to integrate Digital Twin approaches into infrastructure management. By presenting a detailed comparative analysis, it highlights the potential of automated methods as a standardized, scalable, and economically advantageous solution. The insights drawn from this study are expected to contribute towards a paradigm shift, encouraging a balanced combination of manual expertise with the accuracy and efficiency of automated systems. This integrated approach is relevant to the future of road infrastructure maintenance and rehabilitation strategies.