Read AI Summaries carefully and take time and care to edit

[JMS1950]

I’ve done this twice now, mostly as a test.

It seems if there is a main presenter, or if an “MC” introduces each presenter, then the AI Summary can correctly identify who is talking.  But if it’s an ‘open’ group having a discussion, then AI seems to feel that the meeting host is talking unless another person is specifically named.  This was the biggest thing that needed to be edited.

In the second call where it was a social gathering, the AI summary often mixed up who was talking, or assigned a comment to the wrong person.  Yet even here, with no moderator stepping in, it caught a surprisingly large amount of the conversation, even putting things into ‘next steps’.  It summarizes a lot, but groups that under headings.  

I would like to see Zoom give a “thumbs sideways” feedback option — this meeting was neither thumbs up nor thumbs down.

The biggest ‘beware’ is that someone will take and run with the summary without carefully checking it — something that is only possible for someone who was in the actual Zoom meeting.  That’s because the summary sounds so ‘certain’ about what is it reporting.  

Overall, this is quite amazing, and I can see it being potentially very useful.

[CarlaA]

Hi @JMS1950 Thank you for sharing your detailed feedback about the AI summaries! We appreciate your insights regarding the accuracy and attribution during different types of meetings.

Zoom is committed to continuously working on improving the technology to enhance its performance and reliability. I would like to encourage you to submit this feedback directly to our product teams by following the steps outlined here: Sending feedback to Zoom

Thank you again for your thoughtful post! Looking forward to seeing your future contributions 🙂