A working oral-assessment prototype.
CALLIOPE connects assessment design, browser-based spoken dialogue, transcript review, and educator-approved feedback in one pilot workflow.
Educator-provided context
Add public web pages or plain-text sources for use as approved context. Educators verify the resulting questions and scores.
Spoken turn-taking
Browser audio capture and voice activity detection support conversational turns without a separate recording application.
Six draft dimensions
AI evaluators propose scores for Content Accuracy, Explanatory Depth, Scientific Reasoning, Systems Thinking, Clarity, and Metacognition.
Human review closes the loop
Educators inspect the transcript, adjust each draft score, add feedback, and decide what is published to the learner.
Built to learn through co-design.
This pilot is intended for formative use and structured evaluation with educators before any consequential assessment use.
Learners
Explain their thinking aloud and receive feedback only after educator review.
Educators
Review real transcripts and test whether the prompts, rubric dimensions, and workflow support their pedagogy.
Institutions
Evaluate a bounded prototype before deciding what evidence, governance, validation, and integrations a wider pilot needs.