Date
Publisher
arXiv
This study examines whether LLM-generated replies remain semantically consistent when the underlying LLM changes. Using messages from real collaborative conversations, we compared the semantic similarity of generated replies across LLMs under two conditions: with and without preceding chat history. Results show that model choice and conversational context both affect response similarity and alignment with human replies. These findings indicate that prompting and conversational context alone may not be sufficient to preserve response consistency across LLMs, highlighting the need for infrastructure and design strategies that can maintain stable and comparable responses amid the rapid and continuous evolution of LLMs.
Why it matters
Designing conversation-based assessments requires understanding whether LLM-generated replies remain semantically consistent when the underlying LLM changes.
Key findings
- In real collaborative conversations, prompting and chat history alone may not be sufficient to maintain consistent response meanings across different language models.
- The choice of language model and conversational context both affect how similar generated replies are to one another.
- Model selection and conversational context both influence how closely generated replies align with human responses.
Summary drafted with AI from the paper’s abstract and may contain errors. See the full paper for details.
Why use AI?
Study design
