Beyond the Prompt: New Computer Literacy
This preprint addresses a fundamental, unresolved question in modern educational technology: the contrast between the enthusiastic anticipation of LLMs' potential benefits and the thorough scrutiny of the dangers posed by AI-created "hallucinations." The author argues that AI failures in instructional settings are not merely software bugs, but a linguo-pragmatic issue. When users interact with a probabilistic machine using casual, ambiguous natural language, the AI's algorithm dilutes the semantic focus, leading to logical errors. However, within academic problem-solving, natural language transforms into a rigorous, professional sub-language that demands strict structural control. This work is highly relevant for teachers, educational technologists, applied linguists, and curriculum designers looking for a structured, pedagogically secure, and cost-free methodology to safely integrate Generative AI into their disciplines. Fields: Artificial Intelligence in Education, Applied Linguistics, Cognitive Science, Linguistic Pragmatics, Instructional Design.