Video Quality Assessment (VQA) quantifies perceptual video quality by evaluating degradations introduced during acquisition, compression, and transmission processes. The emergence of AI-generated content (AIGC) poses new challenges for VQA, particularly due to its unique distortion types such as synthetic artifacts, geometric inconsistencies, and video-text mismatches which are not well-addressed by existing approaches. Moreover, current methods exhibit limited capability in jointly modeling spatiotemporal dynamics and semantic-level distortions, leading to poor generalization in AIGC scenarios. To address these challenges, we propose a new multimodal video quality assessment (MMVQA) framework, specifically designed for AIGC videos. The proposed MMVQA model integrates technical, aesthetic, and textual information via a dual-path interaction mechanism, with aesthetic features serving as a semantic bridge between visual and textual modalities. This architecture enables comprehensive evaluation of both low-level distortions and high-level inconsistencies. Extensive experiments demonstrate that our method significantly improves objective quality prediction accuracy under complex distortion conditions, offering a novel solution for AIGC video quality assessment.