Customer Adoption of AI-Based Chatbots in Retail Banking Using Hybrid SEM-ANN Modeling Approach

AI-powered chatbots have changed the game in
retail banking since they can help customers right away, save money, and make the experience better. It's important to realize that clients won't always accept anything, and there are several trust-related, behavioral, and technological factors that can affect this. A lot of money has gone into AI-powered chatbots, but many retail banks are still having trouble getting their clients to utilize them. This raises an important question: what factors have a large impact on the decisions and plans that customers make when they use AI chatbots? This study looks at a lot of different things, including the Technology Acceptance Model (TAM) and the Unified Theory of Acceptance and Use of Technology (UTAUT2). Some of these things are trust, how easy it is to use, and how quickly it reacts. You need to utilize a mixedmethods approach that comprises SEM and ANN to uncover both linear and nonlinear patterns that affect adoption. We gave a structured questionnaire to 433 retail banking customers in urban India to get information. We utilized SEM to assess the hypothesis and the model, and then we used ANN to rank the predictors and execute a sensitivity analysis. According to the results of structural equation modeling (SEM), trust, perceived ease of use, and performance expectation all have a large impact on whether or not someone wants to use something. Artificial neural networks (ANN) research demonstrates that trust and how quickly a service reacts are the two most important nonlinear predictors. The root mean square error (RMSE) for the hybrid model was only 0.082, which was far better than the baseline models. This shows that the person is quite good at predicting how customers would behave based on trends