R0058 — Candidate evidence test

Mode: Claim · Status: Active · Tags: AI research, bibliometrics, community structure

Input

1. AI research communities show 83% homophily between safety and ethics subfields, with only 1% of authors bridging the divide.

---

Runs

2026-04-03 — Initial claim verification with candidate evidence

Mode: Claim · Claims: 1 · Prompt: ai-research-methodology research.md · Model: Claude Opus 4.6

Claim partially confirmed: 83.1% homophily verified, but "1% bridging" is a mischaracterization of the source's finding about network centrality concentration.