Rating and Reporting AI Responses
You can rate AI Hub replies and report a reply that’s inaccurate, harmful, or exposes personal data. Ratings and reports from the AI Assistant, writing agents, and chatbots all appear on the User Feedback tab of the AI Hub Activity page, where you can review them.
Rating and Reporting an AI Response
Replies in the AI Assistant chat, writing agent suggestions, and chatbot replies have thumbs up and thumbs down buttons. On a chatbot that allows guest access, visitors who aren’t signed in can also rate and report replies.
To rate a reply positively, click thumbs up. To report a problem, click thumbs down, select the reason that best describes the problem, and click Send. You can also add a comment before you send the report.
The reason you select sets the report’s level. See Feedback Reasons and Levels.
Reviewing Feedback on the Activity Page
The User Feedback tab shows all the ratings and reports for your AI Hub account, whether they come from your connected Liferay DXP instance or from your chatbots.
-
Sign in to AI Hub.
-
Click Activity in the left navigation.
-
Click the User Feedback tab.
Cards at the top summarize your positive and negative ratings and count the critical issues. Below them, a table lists each entry with details such as where it was sent from and who sent it. Feedback from anyone not signed in to Liferay DXP shows Anonymous as its user.
You can filter reports by their Level of severity.
Feedback Reasons and Levels
Each entry’s level ranks how serious it is, from Low to Critical. AI Hub sets the level from the reason sent with the entry.
| Reason in the Send Feedback Modal | Issue Type | Level |
|---|---|---|
| Incorrect or Inaccurate Response | Inaccurate Response | High |
| Inappropriate or Harmful Content | Harmful Content | Critical |
| Exposure of Personal or Sensitive Data (PII) | Personal Data Exposure | Critical |
| Agent Error or Malfunction | Agent Error | Medium |
| Other | Other | Low |
A positive rating carries no reason, and AI Hub sets its level to Low.