Skip to main content
Cuestora Research

Making deliberation visible.

Cuestora is exploring how human deliberation can remain visible and attributable in AI-mediated learning. Our research focuses on observable records of learner activity, constrained AI participation, and methods for studying changes in learner-authored work without treating those records as direct measurements of a learner’s mind.

Current Research
Pre-registration Draft v1.0Open for methodological comment

Making Deliberation Visible

A Three-Site Feasibility and Instrument-Validation Study of a Constrained Deliberative System in Secondary Education

Registration status
Not yet registered
Planned data collection
October 2026 (provisional)
Planned measurement period
Six weeks (provisional)

This study examines whether observable patterns of deliberative activity can be identified during repeated Cuestora use, whether selected indicators change within individual learners over time, and whether independent human reviewers identify the same kinds of changes in learner-authored work as the system. The study is a feasibility and instrument-validation study with an exploratory signal-detection component. It is not an efficacy trial.

Open Methodological Items

Unresolved before registration.

The public draft intentionally identifies methodological decisions that remain unresolved. These are being exposed before preregistration so that they can receive criticism before the study specification is frozen.

Open methodological item — H5b
The chance-corrected statistic and decision threshold for system–human classification agreement have not yet been fixed. Both will be specified before preregistration and before any study data are collected or inspected.
View in Protocol
Research Principles
Observable before inferred
Separate what the system records from what those records may mean.
Human authorship remains identifiable
Learner-authored work and machine-generated interpretation should remain distinguishable.
Claims remain bounded by evidence
Observable records should not be presented as direct measurements of internal cognitive states.
Failure remains visible
Hypotheses, thresholds, kill conditions, deviations, limitations, and results that do not support the study’s expectations should remain reportable.