The fast adoption of generative artificial intelligence (AI) technologies such as ChatGPT in higher education has changed the way of studying, writing, and solving problems, creating serious concerns regarding the AI dependency phenomenon – regular use of AI technologies to replace personal efforts instead of assisting in them. This paper reviews current research to explore whether AI dependency in students is connected with such interrelated outcomes as creativity, problem-solving abilities, and self-regulated learning. Based on meta-analysis, survey, and experiment data from the period from 2023 to 2026, this review reveals that generative AI has an overall positive short-term impact on higher-level thinking skills, with the most significant improvements in problem-solving, moderate improvements in critical thinking, and relatively low improvements in creativity. Nevertheless, the positive effects are very contingent: they are most prominent in students with highly developed SRL skills and deteriorate in cases of prolonged and uncritical use of AI, which is often associated with cognitive offloading, lack of independent thinking, and standardization of ideas. Comparative tables and graphs have been provided that contain summaries of the effect sizes and correlations found in some of the most important papers; finally, the recommendations based on research for effective AI inclusion in learning institutions to ensure cognitive autonomy are offered.