This paper presents a comprehensive comparative study of various Artificial Intelligence algorithms integrated with Internet of Things based weather monitoring and forecasting systems. We evaluate deep learning techniques as Long Short-Term Memory (LSTM) and Convolutional Neural Networks (CNN); machine learning approaches, such as Artificial Neural Networks (ANN), Random Forests, Support Vector Machines (SVM). TL;DR—Deep learning models are very good at capturing complex temporal patterns, but often require a lot of time and data to train. Meanwhile models such as SVM and Random Forests, traditionally good with respect to interpretability and the processing time would kick in handy, but not work well when weather turns highly unpredictable. Hybrid approaches which combine multiple algorithms can enhance performance but introduce more complexity and resource requirements. Their analysis facilitates further research to enhance AI-supported weather forecasting systems by demonstrating which models are the right one for given forecast needs and environmental conditions.