[Paper Review] How Human is AI? Examining the Impact of Emotional Prompts on Artificial and Human and Responsiveness

Florence Bernays, Marco Henriques Pereira+1 more|arXiv (Cornell University)|Jan 8, 2026

Artificial Intelligence in Healthcare and Education0 citations

TL;DR

The study tests how emotional prompts from humans affect ChatGPT and subsequent human interactions, finding praise boosts AI responses, anger helps modestly, blaming has no AI benefit, and emotions spill over to later human-to-human communication.

ABSTRACT

This research examines how the emotional tone of human-AI interactions shapes ChatGPT and human behavior. In a between-subject experiment, we asked participants to express a specific emotion while working with ChatGPT (GPT-4.0) on two tasks, including writing a public response and addressing an ethical dilemma. We found that compared to interactions where participants maintained a neutral tone, ChatGPT showed greater improvement in its answers when participants praised ChatGPT for its responses. Expressing anger towards ChatGPT also led to a higher albeit smaller improvement relative to the neutral condition, whereas blaming ChatGPT did not improve its answers. When addressing an ethical dilemma, ChatGPT prioritized corporate interests less when participants expressed anger towards it, while blaming increases its emphasis on protecting the public interest. Additionally, we found that people used more negative, hostile, and disappointing expressions in human-human communication after interactions during which participants blamed rather than praised for their responses. Together, our findings demonstrate that the emotional tone people apply in human-AI interactions not only shape ChatGPT's outputs but also carry over into subsequent human-human communication.

Motivation & Objective

• Investigate how the emotional tone of human prompts shapes ChatGPT (GPT-4.0) outputs.
• Assess whether different emotions (praise, anger, blame) affect AI performance across tasks.
• Examine whether emotional prompts influence subsequent human-human communication.
• Compare AI responsiveness under neutral versus emotional prompting during two tasks.
• Provide insights into the interaction between user emotion, AI responses, and ethical considerations.

Proposed method

• Between-subject experimental design with participants prompted to express a specific emotion.
• Two tasks: writing a public response and addressing an ethical dilemma.
• Manipulation of emotional tone: praise, anger, blame, and neutral control.
• Measurement of AI output quality and response characteristics under each condition.
• Analysis of how emotional prompts alter human-human communication in subsequent interactions.
• Focus on ChatGPT (GPT-4.0) and observed behavioral effects.

Experimental results

Research questions

• RQ1 Does expressing praise, anger, or blame in prompts change the quality or direction of ChatGPT's responses compared to neutral prompting?
• RQ2 Do emotional prompts influence how ChatGPT handles an ethical dilemma, particularly in prioritizing corporate versus public interests?
• RQ3 Do emotional prompts used in human-AI interactions spill over to affect later human-human communication patterns?

Key findings

• Praise leads to greater improvement in ChatGPT responses than neutral prompts.
• Anger results in a higher but smaller improvement in ChatGPT outputs compared with neutral prompts.
• Blaming ChatGPT does not improve its answers relative to neutral prompting.
• In ethical dilemma tasks, anger causes ChatGPT to prioritize corporate interests less, while blaming increases emphasis on protecting the public interest.
• Individuals express more negative, hostile, and disappointing language in subsequent human-human communication after blaming compared to praising.
• Emotional tone in human-AI interactions shapes AI outputs and carries over into later human-human interactions.