This paper presents an online Human-AI collaborative tool designed to steer a genetic algorithm in solving the Maximum Clique Problem (MCP). By employing human-in-the-loop principles, the system enables users to dynamically adjust algorithmic parameters in real-time. We developed an interactive tool to investigate possible directions for these challenges, enabling an intense collaboration between human experts and artificial intelligence in addressing an NP-Hard problem. We assess the tool by conducting tests with DIMACS benchmark datasets to validate the effectiveness of the approach and gain insights about this form of Human-AI interaction. Additionally, a heuristic evaluation was carried out with domain experts, who interacted with the system to assess usability. Heuristic evaluation confirmed that the tool meets basic requirements, with 17.29% of observations highlighting areas for improvement, particularly in tool functionality and visualization enhancements. The experimental results indicate that the collaborative tool achieves competitive performance while enhancing user engagement and control over the optimization process. Additionally, the results showed interaction preferences, strengths, and weaknesses.