The rapid integration of renewable energy generation has reduced the flexibility regulation capacity of power systems, necessitating the exploration of adjustable resources on the load side to establish a novel ”source-load interaction” balancing mechanism. Air conditioning (AC) load, as a critical demand response resource, has garnered increasing attention. However, existing AC load control strategies are either heavily influenced by user behavior uncertainty or overly reliant on communication and measurement infrastructure. Moreover, most approaches adopt random switching control methods, which fail to maximize user participation willingness and overlook the dynamic variations in user responsiveness, ultimately limiting their practical effectiveness. To address these challenges, this study proposes a dynamic scheduling model that comprehensively considers the aggregated response potential of AC loads and the characteristics of multiple flexible load types, thereby fully exploiting the coordinated regulation capability of load-side resources. Targeting a low-carbon community scenario (incorporating distributed wind power and residential users), the model is formulated with the dual objectives of maximizing wind power accommodation and minimizing source-load power deviation. A greedy algorithm is employed to iteratively solve the maximum available response capacity and actual dispatchable quantity of AC loads in each time slot, enabling dynamic updates of potential assessment and scheduling decisions.