After experimenting with ChatGPT while building the Docsify-This project, I've learned where AI can help and where it can harm. Here are my developing boundaries for when and how to use it effectively.

My (Developing) Guiding Principle

Generative AI (i.e. ChatGPT) can be a powerful efficiency multiplier for open-source work - use it to work smarter on routine tasks and exploration, but never let it replace direct user understanding, code comprehension, or strategic judgment. Project maintainers must always understand what they're building and why.

With that foundation, here's what I've found works and what doesn't:

Leveraging Chatgpt for Open Source Product Design: Building the Docsify-This Project

👍🏼 Do's

• Secondary Market/User Research
• Problem Space Explorations
• Marketing Copy Improvements
• Generating Placeholder Content
• Improving User Interface Microcopy
• Explaining Publicly Available Code
• Suggesting Improvements to Existing Code
• Generating Possible Code Examples

👎🏼 Don'ts

• Replacement for Direct User Research
• Creation of Personas, Scenarios, etc.
• Uploading/analyzing User Research Data (without express permission to do so)
• Use Javascript Code that is not Fully Understood (i.e. explainable to others)