CROSSMMLA

Background & Motivation

AI in education now acts as a learning partner, not just a content source. Students co-write essays with chatbots (Khosravi et al., 2025; Li et al., 2025), pair-program with coding assistants (Lyu et al., 2025), and solve problems in groups where an AI agent contributes ideas (Srivastava et al., 2026). This raises a question for the LAK community: what does productive collaboration between humans and AI look like, and how do we measure it?

Most research on human–AI interaction in education relies on prompts, chat logs and learning outcomes. These capture what learners say to an AI, but miss how the interaction unfolds. MMLA captures high-frequency data across human activity (Blikstein, 2013; Becerra et al., 2026). It can reveal whether learners engage critically with AI output or accept it passively, how a group redistributes talk and roles when an AI joins, when trust turns into over-reliance, and how teachers orchestrate classrooms where AI is present. Research on hybrid intelligence argues that AI should augment human capabilities rather than replace them (Cukurova, 2025); MMLA offers a way to test whether that augmentation happens.

Studying human–AI collaboration with MMLA raises three challenges:

• Conceptual: Shared understanding, joint attention and equal participation were defined for human groups. Do they still apply when one partner is an AI, and what new constructs do we need? (Choi et al., 2026)
• Methodological: Aligning human signals with AI system states needs new data pipelines. Using multimodal AI to analyse these data means AI evaluates collaboration with AI (Fonteles et al., 2026).
• Ethical: Capturing video, voice and biometric data already raises concerns (Alwahaby et al., 2022; Prinsloo, Slade, & Khalil, 2023). Commercial AI systems also raise questions about third-party data sharing, informed consent, bias in how AI supports different learners, and environmental cost.

The CROSSMMLA series has advanced MMLA research for over a decade, bringing scholars and practitioners together around new methods, tools and applications. This edition asks whether AI can be treated as a collaborator, and how MMLA can help us understand, evaluate and design for human–AI collaboration in learning.

Participation & Format

This is not a sit-and-listen event. The half-day workshop welcomes up to 40 participants, from newcomers to experts in multimodal analysis. You need no technical expertise, only an interest in human–AI interaction, collaborative learning, or research design in multimodal learning analytics.

Participants work hands-on with their own dataset or an example dataset provided by the organisers, then reflect together. Bring a short pitch about your research interests and the kinds of multimodal data you work with.

Submit a 500-word abstract on GenAI’s potential to unlock the semantics frontier of MMLA. We welcome:

– Demos & Emerging Tools — Prompt-based pipelines for transcribing, coding, or real-time feedback
– Original Research — How human-GenAI interactions (Tutoring, Co-creating, Coaching, etc.) affect student agency and outcomes.
– Practitioner Reports — Experiences deploying GenAI tools in real classrooms or group settings.
– New Methods & Theories — Frameworks for social knowledge construction, hybrid intelligence, and mitigating over-reliance.

Planned Schedule

Time	Activity
30 min	Elevator Pitches & Framing: Participants introduce their research and data. Facilitators present key definitions and curated examples to frame the day.
90 min	Human–AI Collaboration Research in MMLA: Organisers show how MMLA techniques study human–AI interaction. Participants apply the tools to their own or a provided dataset and share findings.
60 min	Future Considerations: Small groups discuss challenges and opportunities, including data privacy, algorithmic bias, environmental impact, transparency and over-reliance on AI.
30 min	Reflections & Next Steps: A short reflection survey, visualised with text analysis and plots, opens the closing discussion.

Expected Outcomes & Community Relevance

For participants

• Turn collaboration constructs into observable multimodal indicators
• Annotate multimodal data hands-on
• Compare human and AI-generated analyses critically
• Weigh privacy, fairness, transparency and learner agency

For the community

• An initial set of multimodal indicators and constructs for human–AI collaboration
• Methodological and ethical insights for future studies
• New links across learning analytics, CSCL and HCI
• A publication or report setting out a research agenda for MMLA of human–AI interactions and collaboration

tay Connecte

References

Alwahaby, H., Cukurova, M., Papamitsiou, Z., & Giannakos, M. (2022). The evidence of impact and ethical considerations of multimodal learning analytics: A systematic literature review. In The Multimodal Learning Analytics Handbook. Springer, Cham. https://doi.org/10.1007/978-3-031-08076-0_12

Becerra, A., Cobos, R., & Lang, C. (2026). Enhancing online learning by integrating biosensors and multimodal learning analytics for detecting and predicting student behaviour: A review. Behaviour & Information Technology, 45(9), 1885–1910.

Blikstein, P. (2013). Multimodal learning analytics. Proceedings of the Third International Conference on Learning Analytics and Knowledge, 102–106. https://doi.org/10.1145/2460296.2460316

Cukurova, M. (2025). The interplay of learning, analytics and artificial intelligence in education: A vision for hybrid intelligence. British Journal of Educational Technology, 56(2), 469–488.

Fonteles, J. H., Cohn, C., Ayalon, E., Zhou, M., Ts, A., Davalos, E., … & Biswas, G. (2026). Analyzing embodied learning in classroom settings: A human-in-the-loop AI approach for multimodal learning analytics. Learning and Instruction, 103, 102274.

Khosravi, H., Shibani, A., Jovanovic, J., Pardos, Z. A., & Yan, L. (2025). Generative AI and learning analytics: Pushing boundaries, preserving principles. Journal of Learning Analytics, 12(1), 1–11.

Li, X., Li, T., Yan, L., Li, Y., Zhao, L., Raković, M., … & Fan, Y. (2025). An advanced AI-powered engine to facilitate hybrid human-AI regulated learning. Computers & Education, 105527.

Lyu, W., Wang, Y., Sun, Y., & Zhang, Y. (2025). Will your next pair programming partner be human? An empirical evaluation of generative AI as a collaborative teammate in a semester-long classroom setting. Proceedings of the Twelfth ACM Conference on Learning @ Scale, 83–94.

Prinsloo, P., Slade, S., & Khalil, M. (2023). Multimodal learning analytics: In-between student privacy and encroachment. A systematic review. British Journal of Educational Technology, 54(6), 1566–1586. https://doi.org/10.1111/bjet.13373

Srivastava, N., Humburg, M., Burriss, S., Jain, S., Cohn, C., Kim, Y., … & Biswas, G. (2026). The role of LLM-powered conversational agents in supporting inquiry in a narrative-centered learning environment: A learning analytics study. Proceedings of LAK26: 16th International Learning Analytics and Knowledge Conference, 947–954.