User-turn generation reveals that LLMs' interaction awareness is largely decoupled from task accuracy, remaining near zero in deterministic settings even as accuracy scales to 96.8% on GSM8K.
Neural theory-of-mind? on the limits of social intelligence in large lms
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.AI 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Beyond the Assistant Turn: User Turn Generation as a Probe of Interaction Awareness in Language Models
User-turn generation reveals that LLMs' interaction awareness is largely decoupled from task accuracy, remaining near zero in deterministic settings even as accuracy scales to 96.8% on GSM8K.