In the modern classroom, we face a critical pedagogical dilemma: students’ growing reliance on Artificial Intelligence is beginning to replace their reasoning rather than simply providing faster answers. The temptation to default to AI-generated outputs often leads to passive copying instead of active inquiry, creating a barrier to genuine scientific literacy. Even with structured routines in place, many students struggle to meet expectations because they lack the critical-thinking skills necessary to evaluate the accuracy of the tools they use. This session explores a proactive approach to this challenge by integrating AI as a support for, not a replacement of, student thought. To address this, I have developed a specialized assessment tool aligned with scientific thinking using SchoolAI’s Learning Spaces. By collecting student work samples both before and after AI feedback, this project tracks key trends in reasoning depth, the quality of evidence used, and the student's ability to critique AI-generated misconceptions. Combined with student voice data gathered through surveys and discussions, this methodology provides a roadmap for educators to strengthen AI implementation. The goal is to ensure that technology serves as a catalyst for cognitive growth, empowering students to develop the essential scientific reasoning skills required in an increasingly automated world.