Kind Prompting

Kind Prompting is an interactive demo web application that lets users compare how artificial intelligence language models respond to the same request phrased in kind versus rude tones. Built by AE Studio, an applied AI and alignment research firm established in 2016, the tool takes a users text input and generates two versions simultaneously: one reframed as a polite, respectful request and the other as an aggressive, entitled demand. Both versions are sent to the same AI model and their responses are displayed side by side for direct comparison of how tone influences output.

How It Works

Users enter a prompt into a text area, select an AI model (GPT-3.5 or GPT-4), and click Generate. The tool automatically rewrites the input into a kind prompt and a rude prompt using the OpenAI API, then submits both transformed prompts to the selected model. The AI responses for both tones appear simultaneously, each labeled with the rewritten prompt text that was actually sent. The interface highlights that nearly half of Americans believe politeness matters when interacting with AI, framing the comparison in the broader context of human-AI communication norms.

Key Features

• Side-by-side comparison of AI responses to kind versus rude versions of the same user input for immediate visual feedback
• Model selection supporting GPT-3.5 Turbo and GPT-4 with automatic prompt transformation for both tones
• Built-in prompt rewriting that reformulates plain text into respectful and aggressive variants without manual effort
• Free single-page interface built with React that requires no account, login, or installation to use

Use Cases

• Educational demonstrations for students and researchers studying prompt engineering and AI behavioral variation
• Awareness tool for general users exploring how communication tone affects the quality of AI interactions
• Quick experimentation for developers investigating prompt sensitivity and model response differences

Intended Users

Kind Prompting is designed for anyone interested in understanding the relationship between communication tone and AI model output. It requires no technical expertise and is free to use. The tool is especially relevant for educators, students, writers, and technology enthusiasts who want a concrete demonstration of how language framing shapes AI behavior.

Privacy and Security

The tool sends user prompts to external AI APIs (OpenAI) for processing. It does not require accounts, authentication, or personal information, and no user data is stored on the server. Prompts and API responses are transmitted over the internet during use.