A Boundary-Aware Synthesis of Llm Social Agents And Multimodal Executive Agents: Robust Evaluation under Distribution Shift

Abstract

This review examines a shared methodological problem in LLM social agents and multimodal executive agents: how evidence from a realistic benchmark centered on persistent LLM-based social-media agents can be placed in analytical dialogue with a paired text-only and multimodal benchmark for constrained executive decision tasks without erasing differences in scale, assumptions, or intended use. The review draws on two focal records and 12 established sources already present in the project evidence cache. Its comparative framework links behavioral realism, memory, and interaction effects to downstream questions of safety and benchmark validity. Comparison reveals recurring trade-offs among behavioral realism, memory, and interaction effects. These trade-offs do not support a universal ranking; instead, they identify the operating envelope within which each method remains credible and the perturbations most likely to expose fragile conclusions. The article concludes with a research agenda built around transparent comparators, targeted stress tests, and evidence records that can be reused without overstating causal or practical reach.