<![CDATA[Generative AI assistants such as Claude AI are increasingly used by informatics students, yet trust in them depends not only on output quality but also on how much control users feel they have over the process. This study examines the effect of Human Control on trust among Informatics Engineering students at Universitas Malikussaleh using Claude AI. A cross-sectional survey of 55 students measured Human Control through four indicators (controlling AI decisions, correcting AI output, determining AI use, and retaining final decisions) and trust through an eight-item scale (1–5). Both Human Control (M = 3.95) and trust (M = 3.62) were high. The habit of checking AI output (M = 4.07) was stronger than the ability to fix wrong parts oneself (M = 3.76), and consistency of answer quality was the lowest-rated trust item (M = 3.49). Simple linear regression showed that Human Control had a positive and significant effect on trust (B = 0.733, β = 0.742, p < .001), explaining 55.1% of its variance (R² = .551). Perceived control is therefore a strong, but not the only, determinant of trust in generative AI, supporting learning designs that train students to verify and correct AI output.]]>
Copyrights © 2026
