This research introduces StressPrompt, a novel methodology that examines whether LLMs exhibit stress responses similar to humans when under pressure.

• LLMs show performance fluctuations under different stress-inducing prompts
• Researchers created stress prompts based on established psychological frameworks
• Performance patterns reveal interesting parallels between human and AI responses to stress
• Results suggest new approaches for evaluating AI robustness

Medical Implications: Understanding how LLMs respond to stress has significant applications for healthcare systems where AI assists in high-pressure medical environments. This research could help design more resilient AI systems for critical care settings and improve human-AI collaboration during stressful medical scenarios.

StressPrompt: Does Stress Impact Large Language Models and Human Performance Similarly?