Artificial Intelligence based Information Provision in the Automotive Domain

Anticipating and tracking vehicle price variations is vital for the auto vehicle industry. Within this thesis, we trained machine learning models that can predict the price changes based on different events associated with the vehicle models. We divide the price changes into four different categories. The four categories include price decrease, mild price increase, significant price increase or no change at all for a given vehicle model. All predictions are made by taking into consideration the percent pure price change of a vehicle that also takes into account if the overall value of the vehicle changes with price change. We introduce a new feature called inverse premium ratio for every model that indicates the brand value of the vehicle. Inverse premium ratio results as a highly useful attribute to predict the price changes related to vehicle models. In the second part of the thesis, we focused on how to create automatic reports and news on price changes using AI models. We used state-of-the-art generative artificial intelligence models to create news articles that inform the readers about the evolving auto vehicle industry and anticipated price variations. We found several limitations of AI models in terms of logical reasoning and internal biases. We address such limitations by providing clarity and eliminating ambiguity in input data and commands. We introduced a dual agent Reporter-Auditor paradigm to monitor the logical reasoning of AI models in a cost-effective way.