Scaffolding without Execution: How scoped AI features shape planning and reflection in a gamified learning platform
Paper in preparation with Nikodem Płocki, from the full Regula study.
What I did
I led the research: the protocol, the instruments, the analysis in R and the writing, with Nikodem Płocki as co-author. I also designed Regula, the platform the study runs on, and built it with him. The paper works from the full sample, 54 students across two waves, which supersedes the 27 in our thesis.
Where the title comes from
The median AI-enhanced participant did nothing at all in week one.
Everyone had access for all fourteen days. Most activity in both groups came in week two, and AI support didn't pull it any earlier: tasks created in the AI-enhanced group went from a mean of 4.4 in week one to 8.7 in week two. Planning got easier without starting any earlier, and the paper's title comes from that gap.
What we found
System logs
3.3×
As many tasks created by AI-enhanced participants.
Mean 17.2 against 5.2, p < .001, r = −.81.
Interviews
10 of 10
AI-enhanced interviewees described changing how they studied after reflecting, against 7 of 10 Controlled.
Questionnaires
4.58 vs 4.49
Metacognitive self-regulation on the MSLQ, AI-enhanced against Controlled.
Completion rates didn't differ: 80.6 per cent against 70.3, p = .284. Behaviour and interviews moved, and the two validated questionnaires registered none of it. The pilot result, that AI-enhanced participants rated their self-regulation higher than their logs showed, did not survive the full sample, and the paper doesn't claim it.
Where it is
The manuscript is in preparation for PERSUASIVE 2027. None of it has been peer reviewed yet, and nothing on this page should be read as a published finding.