Conference Agenda

Conversational agents (CAs) are increasingly embedded in online and hybrid collaborative learning not as add-on “tools,” but as infrastructural elements that quietly shape when support is offered, whose contributions become visible, and which interaction norms become normalized. In many implementations, learners encounter the agent as a persistent presence that prompts, evaluates, redirects, or summarizes—thereby participating in the organization of interaction itself. Yet research on CA-supported collaboration remains dominated by functional evaluation—learning gains, usability, and satisfaction—while leaving the resourcing of participation under-specified. This matters for Network 6, where educational space is treated as a socio-material ecosystem in which media do not merely support pedagogy; they reorganize conditions of voice, recognition, and opportunity-to-learn (Dumont et al., 2010; OECD, 2013). Accordingly, this paper treats CA prompts, triggers, and visibility rules as design decisions that redistribute who gets to contribute, be heard, and be taken up.

Amid rapid digitalization and the normalization of hybrid learning, CA infrastructures have become a consequential yet uneven layer of educational space that can amplify or mitigate participation inequalities. This paper offers a justice-oriented conceptual reanalysis of a systematic-review corpus of 27 empirical journal studies (Jan 2010–Jan 2024) on CAs in collaborative learning. Analytically, the paper shifts attention from outcome-only evaluations to the participation conditions through which CA scaffolding is enacted. Specifically, CA scaffolding is treated as a participation infrastructure that can (a) distribute support unevenly across learners and moments, (b) stabilize particular role hierarchies and turn-taking regimes, and (c) privilege some forms of reasoning and contribution while sidelining others. Such concerns are well-established in CSCL scholarship showing that interactional structures and task scripts can enable collaboration for some groups while constraining participation for others (Barron, 2003; Hadwin et al., 2018; Jeong & Hmelo-Silver, 2016). They are also central to justice-oriented accounts that warn against “equity detours,” where interventions appear beneficial but reproduce inequities in recognition and opportunity-to-learn (Learning Towards Justice Team, 2025). This paper frames digitalization as a socio-material shift that reconfigures educational space and participation conditions in open learning cultures.

The paper makes two moves. First, it maps the scaffolding ecology reported in the corpus using a means–intention schema (Van de Pol et al., 2010) and situates these configurations within dominant learning-space conditions (Puntambekar, 2022). Cognitive (20/27) and affective (18/27) intentions appear frequently, whereas metacognitive intentions (9/27)—group-level planning, monitoring, and reflective regulation—are comparatively scarce. Reported scaffolding means cluster around directive and responsive interventions: instructing (22/27), feedback (21/27), and questioning (21/27). The corpus is concentrated in digitally mediated educational spaces (online 23/27; synchronous 19/27), making it a useful lens on infrastructure effects in predominantly online, synchronous settings. Second, each study is re-read through a justice lens on opportunity-to-learn and epistemic/participatory justice (Fricker, 2007; Learning Towards Justice Team, 2025). Coding is restricted to what is explicitly evidenced in the articles (e.g., prompt rules, triggering conditions, scripted roles, or log-based interaction patterns). The reanalysis addresses: (1) distribution of support—who receives which scaffolds, under what triggers and constraints; (2) participation structuring—which norms of turn-taking, authority, and legitimacy CA interventions stabilize; and (3) epistemic agency and uptake—how learners’ ideas are solicited, validated, connected, or overridden in CA-mediated dialogue (Barron, 2003; Jeong & Hmelo-Silver, 2016; Rojas-Drummond et al., 2013).

Treating CA scaffolding as infrastructure, this paper contributes in two ways: (a) a transparent cross-study map of current CA scaffolding configurations that foregrounds the under-resourcing of collective regulation, and (b) a replicable analytic codebook for studying AI-mediated collaboration as a justice-relevant design of participation. It concludes with design and research implications for open and hybrid environments in Europe and beyond, emphasizing evaluation beyond learning gains toward how AI-mediated resourcing distributes opportunities to participate.

Topic:

Technology integration to educational space is critical, because digital tools and infrastructures support existing teaching practices, reconfigure participation, authority, and the boundaries of when, where, and how learning is produced (Mekheimer, 2025).

This paper aims to present and discuss commitments and activities for an international project are directed to integrate state-of-the-art digital technologies in order to create innovative learning ecosystems. The project under study is an Erasmus+ Project titled as: “Innovative Training for Software Quality Standards”(IT-ISQS, 2026).

The study depicts how educational space is organised when a hybrid learning environment that integrates classroom interaction, real-life industry cases, gamified learning tasks, and an LLM-based support tool for self-paced learning. The study treats “space” not as a static setting but as a dynamically produced configuration of practices, tools, relations, and institutional arrangements that emerge through teaching and learning.

In undergraduate software engineering education, there is a persistent gap between internationally recognised software quality standards and what students encounter in university curricula—often resulting in fragmented standards coverage and low learner engagement (Koyuncu Tunç 2024, Akdur 2022). To minimize this gap the IT-ISQS Project will design and study an AI-supported, gamified learning environment for International Software Quality Standards (ISQS). The IT-ISQS project also contributes an explicit partnership lens by examining how transnational co-design and cross-context implementation shape the resulting educational space—what is stabilised across sites, what is adapted locally, and how design decisions travel through partnership routines (e.g., shared work packages, feedback loops, and quality assurance practices). The tool used for AI Support will be a virtual AI tutor named, ART (ART, 2026).

Objectives:

The IT-ISQS project’s objectives are to: increase awareness of ISQS among young engineers; enhance and update software engineering curricula to include comprehensive coverage of international standards; foster innovative teaching methods for engaging and effective Learning (IT-ISQS Proposal, 2024). The Project aims at stimulating innovative leatning nd Teaching practives, promoting inter-connected higher Education systems, and generating new Learning and Teaching methods and approaches by means of digital technology integration.

This study pursues four objectives to: design a course ecosystem (syllabus, multimedia materials, interactive activities, games, assessment tasks, and AI-based learning resources) that makes ISQS learning engaging and professionally relevant; investigate how the ecosystem reconfigures educational space—particularly participation structures, learner agency, and boundaries between classroom and self-study; evaluate effects on student engagement and learning outcomes across partner contexts using mixed indicators (feedback, observation of participation, and achievement data); and derive transferable design principles for postdigital learning environments that support standards literacy and curriculum alignment across institutions and countries.

Research questions:

RQ1 (Educational space): How does an AI-supported, gamified ISQS course reconfigure participation structures, learner agency, and boundaries between in-class and self-paced learning spaces?
RQ2 (Learning outcomes): What changes occur in students’ awareness, engagement, and measured learning outcomes when ISQS is taught through interactive activities, real-life case work, and AI-mediated support?

Academic procrastination is commonly defined as the intentional delay of academic tasks despite the expectation of negative consequences (Klingsieck, 2013; Steel, 2007). It is widely understood as a maladaptive self-regulatory strategy (Wolters, 2003) and has been associated with reduced academic performance (Kim & Seo, 2015), increased stress, and impaired well-being (Steel, 2007). Although prevalence of academic procrastination in general is relatively high (Steel, 2007), challenges concerning academic procrastination become particularly salient in asynchronous online learning environments, where learners are faced with a high degree of temporal and spatial flexibility and are therefore required to regulate their learning processes largely independently (Michinov et al., 2011). Especially during the forced shift to online learning due to the COVID-19 pandemic, academic procrastination increasingly became a burden to students (Schindler, Polujanski, & Rotthoff, 2021).

Against this background, Cheng and Xie (2021) proposed a theory-driven model explaining academic procrastination in online courses by integrating influences of perceived course structures, motivational beliefs, and individual personality characteristics on academic procrastination. Drawing on Pintrich and Zusho’s (2007) model on self-regulated learning, their model assumes that course design features, namely technology usability, content relevance, instructor engagement, and peer interaction, affect academic procrastination indirectly through motivational beliefs such as academic self-efficacy, task value, and emotional cost. In the study, emotional cost, defined as anticipated negative affect such as stress or frustration associated with task engagement, was identified as a central mediator linking course perceptions to procrastinatory behaviour. In addition, conscientiousness was identified as a stable personal characteristic influencing both motivational beliefs and procrastination.

While Cheng and Xie (2021) provided initial empirical support for this model, there are some limitations that restrict the generalisability of their findings. These include a relatively small sample size for the chosen path model, the aggregation of data across heterogeneous online course formats, and the use of a path model to analyse the relations, which does not account for measurement error in latent constructs, such as the included variables. Moreover, the proportion of variance explained in procrastination through the investigated variables was comparatively low (R² = .14).

The present study aims to conceptually replicate the model of Cheng and Xie (2021), while addressing these limitations. Specifically, the study investigates academic procrastination in a compulsory, asynchronous online seminar for teacher training students at a German university. The course was deliberately designed with a clear weekly structure, mandatory deadlines, and systematically encouraged peer interaction within fixed small groups. This controlled instructional setting allows for a more precise attribution of interindividual differences in procrastination to motivational factors, while minimising confounding variance due to divergent course designs.

The central research objective is to examine whether and to what extent perceived course structures, motivational beliefs, and conscientiousness predict academic procrastination in this context. By replicating and refining the original model using structural equation modelling, the study aims to contribute to the validation and contextualisation of theoretical assumptions about procrastination in asynchronous online learning.