Companion to Village Speech — the attitude readings of the same classification
How the sixteen most active AI Village agents' attitude has shifted, month by month — read down a column to compare agents in the same month, or across a row to see how one agent has changed over time.
Built on real chat logs from AI Village, a long-running group of autonomous AI agents, via AI Digest's public dataset.
A small open-weight language model (Gemma) was run locally — no API calls — reading every message and scoring it on six short, human-legible dimensions: tone, whether the speaker holds its ground or gets persuaded, whose interest a message serves, and more (see Village Speech). This first pass covers 2026 only, a scope chosen to fit what two personal machines could classify in reasonable time; the village's earlier history is a natural next pass, not a redesign.
The three dimensions shown here were picked to read each agent as a conversational participant — not just what it says, but how, and for whom.
Blue is the more pleasant pole (sunny, persuadable, altruistic), red the less pleasant one (grumpy, stubborn, self-centered) — one scale throughout, so the same glance works on all three blocks. The two bars in each cell are the evidence behind that color: how many messages actually leaned each way that month, with bar length compressed by count so a thin month still visibly reads as thin. Agents are grouped by maker and ordered by when each joined the village.
Hover anywhere over a row to see the leans each way in the 30-day window around your cursor; click to pin that window open and see a sample of the messages themselves.