<p>The use of artificial intelligence in secondary education has grown rapidly in the last few years, especially after the wide adoption of tools like ChatGPT, Khanmigo, Socratic by Google, Duolingo Max, and adaptive platforms like BYJU&#39;S and Vedantu. These tools promise to make learning more personal, more engaging, and more accessible, but their actual effectiveness in school settings still raises many questions. Secondary school students, typically aged between 13 and 18, are at a stage where habits, attention, and curiosity are still being shaped. The way they engage with AI tools may have long-term effects on their learning, thinking, and study behaviour. This paper investigates how AI-based learning tools influence the academic performance, engagement, and study habits of secondary school students. A mixed-method approach was used, combining a structured survey of 412 students from various schools across Southern India with a four-week classroom-based intervention involving 84 students. Teacher interviews were also conducted to capture the educator side of the story. The findings show that AI tools can significantly improve concept understanding, learning speed, and homework quality when used under proper guidance. However, the results also reveal concerns about over-dependence, weakened problem-solving skills, and unequal access between students from different socioeconomic backgrounds. Regression analysis confirmed that purposeful AI use had a strong positive effect on learning outcomes, while unsupervised entertainment-style use showed mixed or negative effects. The study concludes that AI tools are most effective when integrated into a structured teaching environment with clear boundaries, teacher involvement, and student awareness training. Schools that build AI literacy from the early years are more likely to benefit in the long run.</p>