>> Papers, Prototypes & Narrative Science
Tacitus sits at the intersection of conflict resolution, organisational psychology, and advanced AI systems. This space will host white papers, experimental results, and implementation notes on how narrative‑aware AI can help humans find common ground without erasing disagreement.
Deliberative & “Habermas” Machines
We track work on AI‑supported deliberation, including research on “Habermas‑like” systems that structure arguments and surface points of convergence in group debate. Tacitus adapts these ideas to real‑world email corpora, where power asymmetries, hierarchy, and fear of retaliation are part of the data.
Finding Common Ground Across Tribes
Research on bridging systems, deliberative polling, and common‑ground detection informs how Tacitus scores candidate formulations and options. The aim is not to push synthetic consensus, but to identify formulations that opposing sides can live with, given their expressed values and constraints.
From RAG to Conflict Graphs
Standard retrieval‑augmented generation struggles with long, relational conflicts. Tacitus experiments with graph‑based retrieval over claims, narratives, timelines, and policies, and will publish benchmarks and ablation studies as the engine matures.
>> Open Questions
How do you measure “resolution quality” rather than just response speed? What is a fair way to model power and vulnerability in institutional conflicts? How transparent should a conflict graph be to the people whose emails populate it?
Tacitus Lab engages with these questions in dialogue with practitioners and researchers. If you are working on deliberative AI, conflict resolution, or institutional design, we would love to compare notes.