Large language models have begun to attract serious interest from researchers who build agent-based models, according to a paper posted on arXiv. The abstract credits LLMs' increasingly sophisticated natural language understanding and generative abilities for this attention, pointing to a community that has traditionally relied on rule-based agents.
The paper is organised around the modelling cycle, and its title promises to cover both current and possible uses. That framing suggests a practical bent: rather than treating LLMs as a curiosity, the authors appear to be asking where the technology already earns its place and where it might plausibly be applied next.
The arXiv abstract is brief, so this digest draws only on the paper's title and opening lines. The full text presumably expands on the specific stages of the modelling cycle where LLMs fit, which is where the real detail will lie.