Usefulness of generative ABMs for empirical and theoretical ABM paradigms

Ascertain whether and how LLM-based generative agent-based models can meaningfully contribute to (i) empirically calibrated and validated models achieving operational validity and (ii) simple theoretical models designed to elucidate fundamental emergent mechanisms.

Background

The paper describes two established routes by which ABMs achieve operational validity: rigorous empirical calibration/validation for realism, and highly simplified models that reveal general emergent mechanisms. It questions whether generative ABMs, given their black-box complexity and validation challenges, can serve either route effectively.

Resolving this uncertainty is crucial for positioning generative ABMs within the broader methodological landscape of social simulation and for identifying appropriate research designs and evaluation practices for the field.