




































BECE. Newsletter 

Vol.19, No.2, 2025, BECE 1958 

 

 

 

 

Can Human-Machine Collaborative Learning Based 

on Generative Artificial Intelligence Improve Student 

Learning Outcomes? A Meta-Analysis of 20 Experi-

mental and Quasi-Experimental Studies 

By He, W., Zhao, S., Abulaiti, W., Ta, W., & Xu, E. 
Correspondence to: Weigang Ta, Xinjiang Normal University, China, E-mail: 

taweigang@126.com  

S a result of the application of artificial intelligence (AI) in education, 

human-machine collaborative learning emerged as a novel learning 

modality, attracting much attention in academia. The advent of generative AI 

(Gen AI) gave new impetus to this modality. Nevertheless, there are debates 

on the effectiveness of Gen AI-based human-machine collaborative learning. 

This article synthesizes the findings of 20 experimental and quasi-

experimental studies, using the meta-analytical techniques, and examines the 

impact of moderating variables, such as the disciplinary domain, type of 

knowledge, duration of intervention, on the outcomes of Gen AI-based hu-

man-machine collaborative learning. 

Research Findings: 

 Compared with traditional learning methods, Gen AI-based human-

machine collaborative learning is more effective in enhancing student 

outcomes. Specifically, Gen AI’s potent capability of generating content, 

translating language, understanding contexts, and replicating scenarios, 

as well as providing instant feedback in customized interaction, can help 

serve the different needs of various learners and alleviate their cognitive 

anxiety. Also, Gen AI has the potential to assist learners in enacting 

brainstorming-based, human-machine collaborative knowledge genera-

tion by providing simulated scenarios.  

 The analysis of moderating variables reveals that Gen AI-based human-

machine collaborative learning is exceptionally effective in the study of 

procedural knowledge in the domain of social sciences; that its effects 

are less significant in experiments with enduring durations of interven-

tion; that practices like group learning, defining Al’s roles in human-

machine interaction, and adopting the flipped classroom have signifi-

cantly positive impacts on the modality’s outcomes; and that there are 

no significant inter-group differences in the disciplinary domain, type of 

knowledge, and learning pattern. 

A 

mailto:taweigang@126.com


 

Vol.19, No.2, 2025, BECE 1959 

This study demonstrates the positive effects of Gen AI-based human-

machine collaborative learning on student outcomes, providing implications 

for the development of pathways for implementing this modality. Future re-

searchers need to pay more attention to improving the pertinence of the mo-

dality’s design, increasing the weight of communal learning, and enhancing 

the precision of Al’s roles. 

 

 
Source: Open Education Research, 2024; 30(05):101-111. 


