In Artificial Intelligence, numerical optimization is an instantly rising research domain. Swarm Intelligence (SI) and Evolutionary Algorithm (EA) are widely used to answer the problems where the optimal solution is required. Inspired by Seagull’s natural behavior, the Seagull Optimization Algorithm (SOA) is a meta-heuristic, swarm-based intelligent search method. SOA algorithm is a population-based intelligent stochastic search procedure that inherited the manner of seagulls to seek food. In SOA, population initialization is crucial for making rapid progress in a d-dimensional search space. In order to address the issue of premature convergence, this research presents a new variation called the Adaptive Seagull Optimization Algorithm (ASOA). Second, a variety of starting methods have been suggested as ways to enhance seagulls’ propensity for exploratory activity. To improve the diversity and convergence factors, instead of applying the random distribution for the initialization of the population, Qusai-random sequences are used. This paper reveals the state-of-the-art population initialization, and a new SOA variant is introduced using adaptive mutation strategies to prevent local optima. To simulate and validate the results of ASOA and initialization techniques, 8 different benchmark test functions are applied; some are uni-modal, and some are multimodal. The simulation results depict that proposed variant ASOA provides superior results.