Core Questions

When should we use automated intervention rather than a teacher-driven intervention?

What procedures can we use to tell when an intervention should be assigned?

When is stealth intervention superior to an obtrusive intervention? What is the difference?

When we have multiple alternate interventions, how should we select one?

What parameters should we consider when evaluating an automated intervention?

Secondary Questions

How can we avoid "intervention rot", where an intervention becomes less effective over time?

How can we address the fact that student models become less predictive once they are used in intervention?

How can we effectively design when humans respond differently to computers than other humans?