.png){kind=icon}
.png){kind=icon}
Return
Mitigating interpersonal uncertainty in collaborative argumentation: Using LLMs as scaffolds for group cohesion, dialogic behavior and learning performance in CSCL
.png){kind=icon}
PRE
AI
B
DOI：10.1016/j.lmot.2026.102249.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Collaborative argumentation requires learners to navigate inherent epistemic and interpersonal uncertainty. While Computer-Supported Collaborative Learning (CSCL) facilitates deep inquiry, the volatility of argumentative conflicts often triggers anxiety and negative behavioral responses, hindering motivational dynamics. To address these challenges, this study adopts a behavioristic lens to investigate the integration of Large Language Models (LLMs) as learning peers in collaborative argumentation, specifically examining their role in regulating behavioral and emotional uncertainty. Fifty high school students were randomly assigned to either an experimental group (supported by DeepSeek as an active group member) or a control group (unsupported). Quantitative data were analyzed using independent samples and paired samples t-tests (for learning performance and cohesion) and Mann-Whitney U tests (for dialogic behaviors), while qualitative interview data underwent thematic analysis. Results indicate that the presence of the LLM significantly reshaped students' experiences: the experimental group exhibited higher learning performance, more adaptive positive dialogic behaviors, and a reduction in maladaptive negative behaviors associated with conflict anxiety. Furthermore, the LLM fostered stronger group cohesion, suggesting that, when positioned as situated scaffolds, AI peers can facilitate the transformation of potential social threats into opportunities for constructive engagement. Interview analyses further reveal that LLMs function as emotional and behavioral anchors. These findings highlight the potential of AI peers in fostering resilience and optimizing motivational dynamics in volatile learning contexts.
Keywords:
Interpersonal uncertainty
Large language models (LLMs)
Collaborative argumentation
Socio-emotional interaction
Dialogic behavior
Group cohesion
Journal
L
IF:
1.8
Papers:
66
Citations:
1.4K
Organization
Cited Papers