SOURCE-LINKED INTELLIGENCE

Can We Read the Mind of an Audio LLM? A Verbalizable, Multilingual Middle-Layer Workspace

arXiv · AI, language, vision and robotics · article · Aug 25, 2026 · UTC

An audio language model is a black box in a specific way: we see what it says, never what it works out on the way there, and chain-of-thought monitoring helps only if the model writes its reasoning down. Reading a base Qwen3-Omni with a logit lens at the audio-token positions, we find that the answer to a spoken question becomes legible - in words - in the model's middle layers, before it emits any token. Five findings follow. (1) The readout carries concepts in neither the question, the options, nor the model's own transcription: on a clip whose verbatim transcription is empty garbling, it re

Read original source ↗ Open in workspace

recordType

paper

region

Global