DEMHAIC
Multimodal Human-AI Collaboration for Doctoral Education

The Project

 

Doctoral education suffers from widespread problems of dropout and low emotional well-being (which, in Castilla y León alone, can affect thousands of students) for which we do not have scalable solutions, either socio-educational or technological. Recent advances in artificial intelligence (AI) and related disciplines (such as Learning Analytics, LA) could offer scalable and customised solutions to support doctoral students, taking into account the uniqueness of each doctoral process.

 

However, like so many complex problems in which human factors (interpersonal, motivational) play a critical role, AI/LA solutions to these doctoral problems have so far failed, and incur significant ethical risks. To address these risks, we propose to use a human-AI teams approach (in which person and AI collaborate exploiting each actor’s strengths to enhance human capabilities) to support the well-being and persistence of doctoral students. Using a design-based research methodology and a human-centered design approach (in three iterations of increasing complexity and scale), the project will develop and empirically evaluate:

 

• a) structured collaborative person-IA processes to enhance the socioemotional and motivational skills needed to overcome the challenges of a PhD
• b) multimodal software architectures and specific AI models for the ubiquitous support of these human-AI collaborations
• c) design guidelines for human-AI team systems in education.

These project results will not only improve the well-being and employability of doctoral students directly. They will also serve as a blueprint for similar socio-technical systems that use AI in a trustworthy and humane way to improve the well-being of other collectives (e.g. in professional learning).

 

 

Publications
2024

• Chejara, P., Kasepalu, R., Prieto, L., Rodríguez-Triana, M. J., & Ruiz-Calleja, A. (2024). Bringing Collaborative Analytics using Multimodal Data to Masses: Evaluation and Design Guidelines for Developing a MMLA System for Research and Teaching Practices in CSCL. Proceedings of the 14th Learning Analytics and Knowledge Conference, 800–806.https://doi.org/10.1145/3636555.3636877