What researchers studied
Researchers at Qingdao Agricultural University assessed AI-assisted blended instruction in an undergraduate crop-breeding course.
The study covered consecutive student cohorts in 2024 and 2025.
Students received either conventional teaching or instruction that added AI-supported interactive activities to approximately 10% of course time.
Both groups followed the same curriculum, teaching hours and instructor.
Researchers compared examination performance, perceived comprehension pressure, learning self-efficacy and AI usage frequency.
They used effect sizes and regression analyses to examine differences and associations.
The evaluation focused on a specific complex genetics course rather than university teaching generally.
What they found
In 2024, the AI group averaged 80.8 on examinations versus 76.9 in the conventional group.
In 2025, corresponding averages were 80.2 and 75.5.
Reported effect sizes were 0.61 and 0.49, but the examination-score differences did not reach statistical significance.
Students receiving AI-supported lessons reported significantly less pressure understanding difficult course material.
They also reported higher confidence in conducting complex genetic analyses, with a probability value below 0.001.
AI-use frequency was positively associated with final scores in regression analyses.
The reported R-squared values were 0.413 and 0.761 in the two cohorts, but these correlations do not establish independent causal effects.
What it means
AI-supported learning may help students approach demanding genetics concepts with greater confidence.
The clearest evidence here concerns perceived comprehension and self-efficacy, not a demonstrated examination advantage.
Educators should distinguish students feeling better supported from proven improvements in achievement.
The findings provide a useful basis for more rigorous trials of blended teaching.
Important limitation
The study involved a limited sample at one institution and only two cohorts.
The authors recommended larger longitudinal studies with validated assessment instruments.
Potential cohort differences and self-reported outcomes limit generalisation, while the non-significant examination results require caution.
Source Information
Study: Impact of AI-assisted blended teaching on undergraduate learning in crop breeding
Authors: Ming Wang, Fuchao Jiao, Haiyan Zhang, Enying Zhang and Jingtang Chen
Journal / Source: Frontiers in Education
Published: 13 November 2026







