LLM simulations show toxic interactions increase debate time by 25%

Researchers have developed a novel method using Large Language Model (LLM) based Multi-Agent Systems to simulate workplace toxicity and quantify its impact on efficiency. By employing Monte Carlo simulations of adversarial debates, the study measured how 'toxic' prompts affected conversation duration. The findings indicate that incivility increases the number of arguments needed to reach a conclusion by approximately 25%, suggesting a quantifiable 'latency of toxicity' that can serve as a proxy for financial damage. AI

影响 Provides a novel, ethical framework for quantifying the negative impact of toxicity on decision-making processes.

排序理由 Academic paper detailing a new simulation methodology for studying social dynamics.

在 arXiv cs.CL 阅读 →

• Benedikt Mangold
• Large Language Model
• Monte Carlo
• Multi-Agent Systems

• 论文
• 安全

AI 生成摘要 · Google Gemini · 来自 1 个来源。 我们如何撰写摘要 →