Zhenhua Xu, Lijia Yue, Meng Wentao
This 8-week parallel-group experimental study observed the effects of DeepSeek-assisted instruction on sports dance (Latin dance) learning among college physical education students. Forty participants were randomly assigned to an experimental group (DeepSeek assistance, n=20) and a control group (traditional teaching, n=20). The experimental group exhibited significantly higher technical scores, self-assessment scores, and total performance scores than the control group (t-values: 3.45–3.92, P<0.05, Cohen’s d: 1.08–1.24, 95% CI: 2.02–6.98). Physical fitness indicators (double flying swallows, sit-ups, single-leg balance with eyes closed, flat turns) also showed significant between-group differences (all P<0.01, Cohen’s d: 1.01–1.45). In total, 85% of the experimental group reported “very satisfied” or “satisfied” with DeepSeek-assisted teaching. These observational results suggest a potential positive association between DeepSeek use and improved sports dance technical performance and physical fitness, rather than confirming causal effects. This study provides preliminary empirical support for integrating AI-assisted tools into college physical education.