Language learning is a complex process influenced by cognitive, emotional, and cultural factors. This pilot study compares human and Large Language Model (LLM) responses for a survey on language learning challenges. Language model is prompted to list out top 50 challenges while learning a new language. Similarly multiple human candidates were asked to provide list of challenges. Our analysis found only 24% similarity between human and LLM-generated responses, highlighting the AI’s limitations in addressing real human struggles. Although LLMs effectively identify general linguistic difficulties, they fail to capture the key concerns of human learners and unique challenges. These findings suggest that LLMs produce commendable responses with utmost grammatical precision yet distant from human authenticity. We also examine the responses from different sources using different AI models. We train the models specifically trained for classifying two classes: human’s concern and LLM’s concern. To ensure eradicating any biasses occurring due to writing style or grammar errors, we paraphrase human responses via LLM. In our experiments, the AI models efficiently distinguish between LLM’s generated responses and human generated responses, with precision and recall values ranging from 86% to 92%, indicating the exclusiveness of two classes.