I have built check-ins that learners were glad to get, and I have seen dashboards that only made people feel watched. The difference was never the technology. One kind says, I noticed you, how is it going. The other says, I am counting your clicks.
Checking in is not checking on. Support that respects the learner assumes competence and offers help. Surveillance assumes the worst and measures compliance. The choice between them shapes whether a learner comes back.
Autonomy-supportive structure
Structure and autonomy are not opposites. Adults keep going when the structure supports their agency instead of replacing it: real choices inside a clear frame, and someone who notices when they act (Ryan & Deci, 2000). Accountability designed that way feels like care. Designed as monitoring, it feels like distrust, and distrust is a reliable way to lose an adult learner.
Checking in is not checking on. Accountability that respects the learner assumes competence and offers help.
The restart: the gap is data, not debt
Adults leave and come back. Life intervenes, and a gap opens in the course. The punitive design treats that gap as a debt to be paid down with shame. The humane design treats it as data. Something changed, here is the way back in. The learners who returned after long absences taught me to build for the return rather than to punish it, because for most adult learners the return, not the uninterrupted run, is the normal case.
Why AI raises the stakes
AI makes fine-grained surveillance cheap. Every keystroke loggable, every pause flagged. That temptation is worth resisting precisely because it is so easy. The same capability can instead power a gentle nudge at the right moment and an easy restart. Cheap monitoring is not the same thing as good support.
The move
Design one visible, low-stakes way back into a course after a gap, with no penalty attached. Build for the return, because it is coming.