Scholarlink

This paper investigates how large language models can be steered to act more strategically in text-based negotiation settings. Two prompt-based action space designs are compared, namely emotional tone prompts and explicit offer prompts, within a negotiation environment, and outcomes are compared in simulated dialogues. The results show that both approaches improve strategic outcomes compared to a baseline, with tone-based actions yielding higher agreement rates and offer-based actions providing more stable tradeoffs. These findings demonstrate how action space design influences agent behavior, providing insights for deployment of large language models in strategic negotiation scenarios to gain an advantage in, for example, online influence operations.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Strategic Steering of Large Language Models via Game-Theoretic Action Space Optimization

  • 会议论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年

查看全文
  • Samuel Lavebrink,
  • Joel Brynielsson,
  • Mika Cohen,
  • Farzad Kamrani,
  • Christoffer Limér,
  • Madeleine Lindström,
  • Marius Vangeli

摘要

This paper investigates how large language models can be steered to act more strategically in text-based negotiation settings. Two prompt-based action space designs are compared, namely emotional tone prompts and explicit offer prompts, within a negotiation environment, and outcomes are compared in simulated dialogues. The results show that both approaches improve strategic outcomes compared to a baseline, with tone-based actions yielding higher agreement rates and offer-based actions providing more stable tradeoffs. These findings demonstrate how action space design influences agent behavior, providing insights for deployment of large language models in strategic negotiation scenarios to gain an advantage in, for example, online influence operations.