The integration of Large Language Models (LLMs) in education has emerged as a key strategic direction on a global scale. The main purpose of the study is to assess the prompting habits regarding the use of LLM in a mathematical context. We aim to understand how users position themselves regarding the prompt formulation of mathematical problems when using LLM-based tools. An evaluation of the impact of prompt formulation is performed on perception of AI output accuracy, user satisfaction, and perceived usability. The findings underscore the critical role of prompt clarity in shaping user satisfaction and AI accuracy, while also revealing that mathematical proficiency is not a sufficient predictor of effective prompting. The diversity of challenges faced by users reinforces the importance of adaptive interfaces and targeted educational interventions to bridge the gap between AI capabilities and user needs.