The integration of AI assistants into customer service has significantly enhanced operational efficiency and customer satisfaction. However, concerns about long-term impacts on employee skill retention, particularly in complex and adaptive tasks, have emerged. This study proposes a six-month longitudinal field experiment involving 300 customer service representatives, examining performance metrics and psychological factors across full AI assistance, partial AI assistance, and a control group. Additionally, an intervention using "AI-free" days and gamified problem-solving exercises is tested to mitigate potential skill erosion. Results will provide empirical insights into balancing AI-driven efficiency gains with the preservation of essential human cognitive skills, guiding responsible AI implementation strategies.