Large Language Models Do Not Always Need Readable Language

2026-06-18

AI Summary

This study investigates whether large language models must rely on human-readable natural language for interaction, particularly in inter-model communication scenarios. To address this, the authors propose BabelTele, a task-agnostic, model-native representation method that enables efficient model-to-model communication through high-density, low-readability, yet semantically recoverable non-standard text. The work provides the first systematic evidence that human readability and model-based semantic recoverability can be partially decoupled. This claim is substantiated through a comprehensive evaluation encompassing readability diagnostics, model likelihood assessments, human surveys, and downstream task performance. Experiments demonstrate that BabelTele achieves a 27.9% reduction in textual volume while preserving 99.5% semantic fidelity, significantly lowering context overhead without compromising performance in cross-model transmission, agent memory retention, or multi-agent collaboration.

Abstract

Large language models (LLMs) are commonly prompted and interfaced with human-readable natural language, even when the intended reader is another model. This paper investigates whether semantic information can be encoded in compact, non-standard textual forms that sacrifice human readability while remaining recoverable by LLMs. We refer to this class of model-centric textual representations as BabelTele, approached here not as a fixed protocol but as an empirical probe into LLMs' capacity to generate and interpret such representations. Through readability diagnostics, model likelihood measures, human questionnaires, and downstream task evaluations, we find that BabelTele can substantially depart from ordinary natural language while preserving core semantics for instruction-tuned LLMs. As a task-agnostic representational paradigm, BabelTele demonstrates high information density, maintaining 99.5% semantic fidelity even when the text volume is condensed to 27.9% of its original length. We further evaluate its semantic robustness in cross-model transfer, agent memory, and multi-agent communication. Results suggest that BabelTele can reduce context overhead while generally maintaining reliable downstream performance, although its effectiveness depends on the compressor-reader pair and task setting. These findings indicate that human readability, natural-language typicality, and model-side semantic recoverability can be partially decoupled, opening a path toward model-native representations in future exploration of LLM systems.

Problem

• Large Language Models
• Human Readability
• Semantic Representation
• Model-Centric Communication
• Non-Standard Text

Innovation

• BabelTele
• semantic compression
• model-native representation
• non-readable language
• cross-model communication