Artificial intelligence in professionally oriented physics education: patterns of use and perceived learning support among future engineers

Artificial intelligence (AI) is increasingly used in engineering education, yet evidence remains limited regarding students' AI-use practices and perceived learning support in professionally oriented physics education. This study aimed to describe AI-use patterns and examine their associations with students' perceptions of professionally oriented physics learning. Methods A quantitative cross-sectional anonymous online survey included 209 students from engineering-related programmes recruited from six universities in Kazakhstan. The questionnaire assessed the frequency and purposes of AI use and six separate learning-related perceptions. Descriptive statistics and Spearman rank correlations, with Holm-adjusted p -values and bootstrap confidence intervals, were used; valid sample sizes varied by item. Results Among 208 valid responses on general AI use, 36.1% reported regular use and 31.7% occasional use. In specific activities, reported AI use ranged from 43.5% for written assignments to 55.1% for clarifying difficult topics. Perceived support was reported by 61.1% for engineering thinking, 42.5% for applying physical laws, and 42.8% for interpreting computational and experimental data. More frequent AI use in physics and specialised disciplines was associated with more favourable responses across six perception items (Spearman's ρ = 0.451–0.605; Holm-adjusted p < 0.001 for all six comparisons). In a sensitivity analysis restricted to AI users, all six associations remained statistically significant, with smaller coefficients ( ρ = 0.234–0.490; Holm-adjusted p ≤ 0.0027). Discussion and conclusion Reported AI use and perceptions varied by activity and outcome. The observed correlations reflect self-reported associations rather than effects of AI on objectively measured competencies. Cross-sectional design, voluntary recruitment, and shared-method measurement limit causal and population-level interpretation. The findings may help instructors identify activities in which students perceive AI support, but recommendations for assessment or instructional design require further evaluation.