Suggested Replies EAP - Reporting on Usage and Helping to Train the AI

We currently have no way to report on when the Suggested Replies feature is used to create a response to a customer, which prevents us from knowing when the AI has recommended a bad macro or when an agent has incorrectly used a feature. In tandem with this issue, we have no way to flag macros as correct/incorrect for a given scenario within the tool. As a result, we have users expressing a lack of confidence in the AI's ability to predict and learn, as well as management hesitation to use the tool due to lack of reporting/accountability. 

We ask for three things to help improve this feature: 

• Reporting capability, especially via the “Events” view within a ticket, to show that Suggested Replies was used. 
• The ability to tell the system “No, this is not the correct macro to use here”
  • Can benefit the system by helping train the AI
  • Will prevent the macro from being repopulated should a user clear the text in the reply. 
• The ability to mark certain macros as “Do Not Recommend” or “Do Not Use” within the Admin Center to prevent special case macros from being recommended. 

This is a regular occurrence, with multiple teams within our organization reporting poor recommendations from the tool,  and managers always ask for reporting and accountability tools in our organization. 

We know of no workarounds for this issue, affected teams just don't use the tool.