New systems of learning for distance learning institutions? A six-study review of implementing AIDA

Synthesis: Rienties et al. (2026) examine how the Open University (UK) — a large-scale distance learning institution teaching 200K+ learners across 50+ countries — designed, implemented, and evaluated an AI digital assistant (AIDA) using Sharples' embedded systems approach. Through six iterative Design-Based Research (DBR) studies over 18 months involving 498 students and 20 staff, they found that purpose-built GenAI tools embedded within the learning environment can enhance engagement in distance education, provided development is participatory, governance is robust, and integration aligns with institutional strategy. In an exploratory randomized controlled trial, students using AIDA spent twice as long and visited more pages in the course relative to the control group.

Key Findings

• Early studies (1–3) gathered student preferences for AIDA's design and identified the benefits of 24/7, context-specific support embedded within the learning environment, alongside concerns over Ethics, academic integrity, and data privacy; around 20% of students were initially highly skeptical about whether the OU should develop and use AIDA.
• Later studies (4–6) explored actual hands-on use. Acceptance increased post-use, especially among initially skeptical participants, with 96% of students expressing interest in having AIDA available in their formal studies — hands-on use of AIDA's quiz, explain, and chat functions shifted perceptions from abstract potential to tangible usefulness.
• In an exploratory randomized controlled trial (Study 6; 115 experimental, 48 control students in an OpenLearn Create course), students supported by AIDA had twice the usage time and visited more pages relative to the control group, though no significant differences were found on overall time and other learning process data — possibly due to the exploratory task design and the sandbox environment differing from the main teaching environment.
• Mapping the six studies against Sharples' embedded systems framework identified enabling factors — senior leadership sponsorship, cross-unit collaboration (between KMi and IET R&D units, Digital Services, and Faculties), iterative DBR refinement, and data-informed decision-making — while also revealing gaps in building systems-thinking capacity and institutionalizing systemic change (early AIDA work operated in "pilot project" mode).
• The paper warns against naive GenAI adoption, citing the demise of KhanMigo (learners were not actually engaging with the chatbot, with limited evidence of positive gains), underscoring that technical capability must be coupled with organizational readiness and governance.

Study Design & Method

This is a six-study, 18-month Design-Based Research (DBR) program conducted at the Open University, UK, involving 498 students and 20 staff. The research is framed by Sharples' (2025) embedded systems approach, which adapts Bronfenbrenner's (1979) Ecological Systems Theory to distance learning — recognizing that how learners engage with AI is influenced by multiple nested layers (peers, materials, tutors, student services, IT support, quality assurance, government AI Regulation in Education, tech companies). It is also informed by the Diffusion of Innovations Theory (Rogers; Jin et al., 2025). Studies 1–3 gathered preferences and perceptions of AIDA's design; Studies 4–6 explored actual hands-on use by students and staff, culminating in Study 6's exploratory randomized controlled trial in an OpenLearn Create course. Findings were mapped against the nine key actions of Sharples' embedded systems framework.

What this means for practice

• Instructional designers. Co-design the assistant with students and staff instead of procuring one: the six studies began by gathering student preferences, and acceptance rose after hands-on use, with 96% of students then wanting AIDA in their formal studies and initial skepticism eroding with experience rather than with persuasion.
• Instructional designers. Embed the assistant inside the learning environment and tune it to your own curriculum — AIDA offered quiz, explain, and chat functions within the VLE — because the paper's warning case is a well-funded generic external chatbot (KhanMigo) whose learners were not actually engaging with it and where evidence of positive gains was limited.
• Institutions. Budget for governance and organizational readiness alongside the technology: senior leadership sponsorship, cross-unit collaboration, iterative DBR refinement, and data-informed decision-making were the enabling factors, while the mapping found gaps in systems-thinking capacity and in institutionalizing change beyond recurring "pilot project" mode.
• Institutions. Judge adoption by process data with realistic expectations: AIDA-supported students spent on average roughly twice as long in the course as the 48 control students (75m59 vs. 30.85 min), but the difference was not statistically significant (Z = −1.433), so an engagement signal is not by itself evidence of learning.
• Educators. Design explicitly for the distance-learning conditions the paper describes — 70% of students already in work and no feasible in-person invigilation — by pairing AI support with visible academic-integrity and data-privacy provisions, since ethics, integrity, and privacy were the concerns students raised most before they had used the tool.

Limitations

The RCT (Study 6) was exploratory: students were asked to explore the OpenLearn Create course at their leisure with no credits or external reward, and the sandbox environment differed from the main teaching environment, which may explain the lack of significant differences in process/outcome data. The study focuses on one institution (the Open University), and the AIDA application is institution-specific. The authors acknowledge that early AIDA work operated in "pilot project" mode with limited systems-thinking capacity, and the mapping to Sharples' framework involves interpretive judgment.

Citation

Rienties, B., Coughlan, T., Domingue, J., & Herodotou, C. (2026). New systems of learning for distance learning institutions? A six-study review of implementing AIDA. Computers and Education: Artificial Intelligence.