GCRD Insights Panel recording
Measuring Democratic Health from the Classroom
Faculty and student researchers from Pakistan, Albania, Greece and the United States discuss what studying democracy becomes when the students are also the ones measuring it.
Monday, 10 August 2026 · Virtual panel · Hosted by AltLiberalArts in partnership with GCRD
If the player does not open, watch it on YouTube.
How can universities prepare students not simply to study democracy, but to take part in its renewal? On 10 August, GCRD and AltLiberalArts brought together the faculty and students already answering that question in classrooms on four continents.
Across six universities, students are using the Democracy Discourse Index to read real public conversation, including comments on their own national news platforms, and to score it for empathy, civility, trust language and democratic agency. Under faculty supervision they do the coding themselves.
The distinction matters. These students are not only learning about democratic erosion. They are learning, through practice, to recognise its early signs in their own public spheres.
"The DDI treats students as knowledge producers, not just students of democracy. Their coding doesn't just teach them about civic life, it becomes part of the infrastructure we use to understand democratic health at scale. That is what makes this classroom model different."
Dr. Jacob Udo-Udo Jacob, Executive Director, GCRD
From classroom coding to national scores
Every contribution in the validation set is coded independently by three student coders, and disagreements are adjudicated rather than averaged away. That triple-coded material is the ground truth that trains ReHuman, GCRD’s AI coding system, to apply the same instrument across corpora far larger than any classroom could read, and to produce topic and national scores. The classroom is not a teaching exercise sitting beside the research. It is the part of the infrastructure everything downstream rests on.
- Triple coding Three trained student coders score every contribution independently, under faculty supervision.
- Adjudication Disagreements go to trained adjudicators, and recurring ambiguity feeds back into the rubric.
- Ground truth The adjudicated set becomes the validated reference against which machine coding is measured.
- ReHuman at scale GCRD’s ReHuman AI system applies the same instrument at scale to produce national scores.
ReHuman’s licence to scale rests on its continuing agreement with that human-coded ground truth. This is why the coding work in the classroom never stops.
Three movements
-
I
The case for measuring discourse
Why democratic health can be read in everyday public language, and the human and AI methodology behind the index.
-
II
Inside the classroom
How the DDI is taught, how student coders are trained, and how intercoder reliability and multilingual adaptation are managed across sites.
-
III
First findings
Early evidence from the founding pilot countries, and a comparative reflection on democratic stress and resilience across societies.
Who is on the panel
Faculty principal investigators
- Prof. Wajiha Raza Rizvi NED University of Engineering and Technology, Karachi, Pakistan Embedded in coursework
- Prof. Juliana Cyfeku Fan S. Noli University of Korçë (UNIKO), Albania Co-curricular certificate
- Prof. Verse Shom Tulane University, United States Embedded in coursework
- Prof. Panagiotis Paschalidis Aristotle University of Thessaloniki, Greece Embedded in coursework
Student researchers
- Redina Skënderi Fan S. Noli University, Korçë, Albania
- Subulna Imran NED University of Engineering and Technology, Pakistan
Why this matters beyond the classroom
Existing democracy indices largely measure institutional arrangements and expert assessments. The DDI measures something adjacent, and easy to miss: the moral, epistemic and relational quality of everyday public language. When discourse turns hostile, conspiratorial or dehumanising, institutions can still look stable while democratic culture is already weakening underneath them. The DDI is built to catch that earlier.
A national score orients the reader, but it is not the whole of the instrument. The decision-relevant object is a topic, in a place, over a period. Trust and agency move differently across issues, and two topics can share the same average while presenting opposite democratic conditions. Reading the configuration beneath the number is what turns measurement into diagnosis, and it is what lets the question shift from what message will move an audience to which democratic condition has to be strengthened before engagement can be productive at all.
It is also a working answer to a question many universities are asking now: how to bring frontier AI into the classroom as a partner in civic knowledge production, governed throughout by human interpretation, local expertise and civic purpose, rather than as a threat to academic integrity or to student learning.
Join the consortium
The DDI research and teaching consortium is growing. Universities interested in bringing the DDI into their own coursework or co-curricular programmes are invited to be in touch.
Register your interest