A Framework for Context Enrichment and Selective Robustness in LLM Social Agents
arXiv cs.CL · 5 Nov 2026, 04:58 AEST
Researchers are exploring the use of large language models as agents in social simulations, focusing on how context enrichment can influence their behavior. The study introduces a framework for controlled context enrichment and selective robustness to improve agent performance.
Key facts
- Researchers use large language models as agents in social simulations, where agents are prompted with observations and background information intended to shape their actions.
- One plausible hypothesis is that this brittleness reflects a lack of grounding and can be reduced by enriching the context provided to the agents.
- We address the first need with controlled context enrichment, a set of prompt substitution rules that systematically enriches distinct types of context: neutral padding, generic social background, descriptive decision grounding, and normative decision grounding.
- We demonstrate the framework in a Prisoner's Dilemma-like game, showing how it can be used to identify which forms of context enrichment produce more desirable and robust agent behavior.
Read aloud
Researchers are exploring the use of large language models as agents in social simulations, focusing on how context enrichment can influence their behavior. The study introduces a framework for controlled context enrichment and selective robustness to improve agent performance. Researchers use large language models as agents in social simulations, where agents are prompted with observations and background information intended to shape their actions. One plausible hypothesis is that this brittleness reflects a lack of grounding and can be reduced by enriching the context provided to the agents. We address the first need with controlled context enrichment, a set of prompt substitution rules that systematically enriches distinct types of context: neutral padding, generic social background, descriptive decision grounding, and normative decision grounding. We demonstrate the framework in a Prisoner's Dilemma-like game, showing how it can be used to identify which forms of context enrichment produce more desirable and robust agent behavior.
Sources in this cluster
- A Framework for Context Enrichment and Selective Robustness in LLM Social Agents
arXiv cs.CL · 5 Nov 2026, 04:58 AEST