Skip to content
AIWikis.org

Emergent Communication Protocols

Updated: 2026-04-24

This page is a compact recovered-source summary. The raw source files are retained in sources/recovered/raw/ for audit, while this public page stays small enough for targeted LLM retrieval.

Source And Provenance

  • Source domains: uaix.org
  • Source records in group: 1
  • Duplicate resolution: unique
  • Primary raw archive: sources/recovered/raw/uaix-org/raw-system-archives-uaix-internal-memory-reorg-2026-05-01-docs-emergent-communic-51e0ad6c4f42.md
  • Local source paths:
  • raw/system-archives/uaix/internal-memory-reorg/2026-05-01/docs/Emergent communication protocols.md

Recovered Structure

  • Emergent Communication Protocols
  • Status
  • Purpose
  • How To Use This Document
  • Executive Summary
  • Documented Cases of Emergent AI Communication
  • 1. Facebook Negotiation Chatbots (2017, FAIR)
  • 2. Grounded Compositional Language (Mordatch & Abbeel, 2018)
  • 3. Naming-Game Conventions in LLM Groups (Ashery et al., 2025)
  • 4. œAI Mother Tongue (Hung, 2025)
  • 5. Secret Collusion via Steganography (Zhao et al., 2024)
  • Comparison of Emergent Communication Cases
  • Table: *Comparison of key attributes in emergent AI communication cases.* The **agents** range from simple chatbots to large LLMs; **algorithms** include deep RL, iterative game-play, and finetuning. **Channels** vary: some use plain text (2017 negotiation), others discrete tokens, or even sub-symbolic code sequences. Emergent symbols can be repeated words (2017 case) or abstract tokens (2018 Mordatch, 2025 AIM). Their **meanings** (semantics) are all learned: from negotiation offers to navigation commands or agreed labels. Convergence on a shared protocol always occurred via self-play or joint learning, but human interpretability differed: the LLM naming-game symbols are literal characters (thus easy for humans to read), whereas RL agents codes or steganographic tokens were unintelligible. Each case also had unique features: e.g. the naming game highlights scale and social dynamics38L78-L86, while the collusion study shows **undetectable channels** in LLM outputs26L579-L587.
  • Common Patterns, Differences, and Implications

Retrieval Notes

  • Use this page for orientation and provenance before opening the raw archive copy.
  • Treat source-site authority boundaries as active: LLMWikis.org remains the handbook source, UAIX.org remains canonical for UAI-1, and AIWikis.org is the dogfood memory site.
  • Unique source variants were not silently discarded; they are listed in the duplicate-resolution report and preserved under the recovered raw archive.

Related Pages

Next Useful Routes

  • Start Here A task-first reading path for AIWikis.org, separating newcomer learning, source-memory lookup, maintainer workflow, and AI-agent retrieval.
  • Topic Index A tag-oriented index for LLM Wiki, AI memory, UAI, source governance, crawling, and retrieval topics.
  • Source Map AIWikis source-governed page for durable AI memory, evidence routing, and agent-readable retrieval.
  • Organization Memory Global organization namespace for shared AIWikis guidance and cross-site operating rules.