Can AI be a good counselor? Comparing the effectiveness of AI and human career counseling

While generative Artificial Intelligence (AI) is increasingly used for career guidance, empirical comparisons with human counseling remain scarce, leaving practitioners without evidence-based guidance for integration. We conducted a three-week controlled trial with 183 Chinese university students, comparing AI and human-led brief career counseling across multiple outcome dimensions: information acquisition, career decision self-efficacy, and decision-making efficiency. Results reveal that the two approaches differ consistently in their effectiveness. AI counseling was non-inferior to human counseling in reducing information deficits, with steeper initial gains. However, human counseling was more effective in enhancing career decision self-efficacy and improving decision-making outcomes, as evidenced by reduced decision time and difficulty. These findings suggest AI excels in information delivery but is less effective in fostering self-efficacy and decision efficiency, positioning it as a complement to human expertise. Practically, our results provide an empirical basis for hybrid counseling models in which AI addresses initial information needs while human counselors focus on sense-making and confidence-building, offering practical guidance for institutions seeking to improve their career support services.