Question Assistant is an AI feature that evaluates question drafts and provides real-time suggestions for improvement before they’re submitted either for human review in Staging Ground, or for publication on Stack Overflow.
How does it work?
Question Assistant uses machine learning models to evaluate askers’ question drafts following this process:
- Askers begin their question draft via the ‘Ask Question’ button by writing a title and describing their problem.
- Machine learning predicts the presence or absence of four categories in the draft (Problem Definition, Attempt Details, Error Details, Missing Minimum Reproducible Example).
- Information is passed to Google Gemini, which provides real-time suggestions on how the asker can improve their draft in the right sidebar.
- Askers can choose to implement the suggestions or ignore them. They can also click “refresh” to get a new set of suggestions.
- Askers then submit their question which will either be published on Stack Overflow or, for some new askers, sent to Staging Ground for human review first.
Why use it?
Question Assistant is designed to help questions be more successful on Stack Overflow. Research indicates that incorporating the suggestions offered has a positive impact on question success. In experiments users who received suggestions from Question Assistant were more likely to get upvotes on their questions and receive answers, than those that did not.
What kind of suggestions will it give me?
All suggestions provided to askers will be within four categories:
- Problem Definition - The problem or goal is lacking information to understand what the user is trying to accomplish.
- Attempt Details - The question needs additional information on what you have tried and the relevant code.
- Error Details - The question needs additional information on error messages and debugging logs (as relevant).
- Missing Minimum Reproducible Example (MRE) - Missing a minimal reproducible example using a portion of your code that reproduces the problem. For more information, refer to this help center article on creating an MRE.
The tool will detect the presence or absence of these categories in the question draft, and will return suggestions for improvement that are informed by details of the specific question you are asking.