Improve the shared AI feedback modal

Context

In #1612 (closed) I proposed a standard mechanism to allow users to give feedback on AI content. The original intention was that we would have a shared, generic feedback form that all AI features would use. The questions and content of the modal would be the same across different AI features.

Problem

We have implemented the generic feedback form for the chat.

The feedback we're receiving is not as actionable as we would like. Based on the feedback we're receiving, we think users assume we can see the question and answer the have in the chat, but we cannot due to legal/privacy reasons.

We would also like the answers to be more actionable so we can further improve the chat. What we would like to know:

• What was the user asking?
• What answer did they receive?
• What answer would they like to receive?

Suggested solution

Extend the shared modal component to allow for a degree of customization between different AI features.

The modal starts in a default, generic state. However, the component has properties on it that allow individual product teams to customize text and add/change questions in order to get the most actionable feedback possible.

Image	Description
	Adds an additional alert that GitLab staff cannot actually see AI content
	This is the default modal but some questions can be overridden. Also additional questions can be added.
	Default versus overrides (chat example)