<![CDATA[This systematic literature review aims to analyze the role of Artificial Intelligence (AI) as a stimulus for students' self-regulated learning (SRL) and mathematical problem-solving abilities, given the persistent gap between the demand for 21st-century problem-solving skills and traditional instructional methods that make students passive. The study employs a Systematic Literature Review (SLR) following PRISMA guidelines, with article searches conducted on Google Scholar using Publish or Perish software and keywords combining "Artificial Intelligence," "Self-Regulated Learning," and "Mathematical Problem Solving." From an initial 200 articles, 15 were selected based on inclusion criteria, including publication years 2024-2026, full-text availability, and focus on at least two of the three main variables. The results indicate that AI significantly enhances SRL by enabling personalized, adaptive learning and providing instant feedback, visualizations, and step-by-step guidance, thereby improving students' initiative and problem-solving skills. However, the findings also reveal critical gaps: without metacognitive mediation, AI risks becoming an "answer provider" rather than a "thinking partner," leading to overreliance. Additionally, most studies lack rigorous designs (e.g., RCT), have limited sample sizes, neglect teacher readiness and infrastructure constraints, and rarely examine specific SRL dimensions such as metacognition or help-seeking behavior. AI tools like ChatGPT, Gemini, Mathos AI, and Photomath dominate the literature, yet valid and reliable instruments to measure AI-assisted problem-solving behavior remain underdeveloped. This review concludes that while AI holds transformative potential, its effectiveness depends on integrating metacognitive strategies, strengthening teacher digital literacy, improving infrastructure, and conducting longitudinal or experimental studies to establish causal evidence. ]]>
