Enhancing Service Innovation Through Ai-Based Prediction Models in Digital Transformation

Abstract

Service innovation has dramatically changed digital transformation by integrating artificial intelligence (AI) based prediction models. This study explores a wide range of studies and computational experiments to determine the effect produced by AI-EfficientNet, DiCENet, LSTM, and transformer-based architecture on predictive accuracy and orientation of the service. The results indicate that transformer-based models surpass traditional machine learning methods with an accuracy of 96.2% and a minimum mean squared error (MSE) of 0.007. The most important features in service innovation are the customer purchase history and real-time demand data. Fortunately, there remains a computational inefficiency problem, and thus, an optimal necessity still exists for model pruning and quantization. This highlights the interdependent role of responsible deployment of AI in the context of the tradeoff between accuracy and ethical and computational concerns. This study provides useful ideas to businesses and politicians on adopting AI and its applications in enhancing decision-making and improving the efficiency of providing a service in the context of digital transformation.