The Use of Artificial Intelligence in Computer Science Lessons and Its Pedagogical Advantages

The article examines the didactic potential, effectiveness and risks of using artificial intelligence (AI) tools in teaching computer science in secondary and higher education. Following the IMRAD structure, a systematic-synthetic review was conducted, comparing international empirical studies and meta-analyses on intelligent tutoring systems (ITS), large language models (LLMs) and code-generation tools. The findings indicate that ITS yield moderate-to-large gains compared with conventional classroom instruction (Cohen's d ≈ 0.66–0.76); LLM-based assistants improve task completion speed and quality among novice programmers, yet, without pedagogical guardrails, may impair independent knowledge acquisition. A five-stage didactic model of an AI-integrated computer science lesson is proposed. The results are of practical value for computer science teachers, curriculum designers and education policymakers.