Causal Inference for Human-Language Model Collaboration

arXiv:2404.00207v1 Announce Type: cross
Abstract: In this paper, we examine the collaborative dynamics between humans and language models (LMs), where the interactions typically involve LMs proposing text segments and humans editing or responding to these proposals. Productive engagement with LMs in such scenarios necessitates that humans discern effective text-based interaction strategies, such as editing and response styles, from historical human-LM interactions. This objective is inherently causal, driven by the counterfactual `what-if' question: how would the outcome of collaboration change if …

abstract arxiv causal causal inference collaboration collaborative cs.ai cs.cl cs.lg dynamics editing engagement human humans inference interactions language language model language models lms paper productive proposals strategies text type