This study examines whether high-school students’ attitudes toward AI-assisted English learning (AI-AEL) shape intentions to choose language-related majors. Grounded in Self-Determination Theory and Social Cognitive Theory, we propose a dual-pathway account whereby positive AI-AEL experiences enhance autonomy and self-efficacy, promoting academic intentions, while negative experiences exert weaker effects. Using purposive sampling, Grade-12 students at a key public high school in Hebei, China completed a content-validated attitude scale, autonomy and self-efficacy measures, and a major-intention item. Reliability and factorability were established; exploratory factor analysis supported a two-factor attitude structure. Regression models with demographic and usage controls showed that positive attitude significantly predicted intention to pursue language-related majors, whereas negative attitude was non-significant. Pedagogically, findings support gradual AI integration, choice-based tasks, micro-feedback cycles, collaborative critique of AI outputs, and activities that convert concerns about dependency, accuracy, verbosity, and authenticity into explicit learning goals. While the study provides novel empirical evidence linking AI-assisted learning attitudes to academic intentions through motivational pathways, limitations include the single-school sample (limiting regional generalizability), cross-sectional design (precluding causal inference), and reliance on self-reported attitudes. Future multi-site, longitudinal studies with behavioral measures are needed to validate the proposed framework and assess long-term impacts of AI-assisted learning on actual major enrollment. Overall, thoughtfully embedded AI can strengthen motivation and self-efficacy, inform students’ academic choices and support ongoing AI and Humanities integration in language education.