Although I try to shorten bullet points as much as possible, I also would like an audience member to be able to scan the slide in a few seconds to remind himself of what I have talked about. To achieve this, is it better to....
Highlighting words is OK, so I'd suggest to go with first option.
While you'd test with real users, in my personal experience I noticed that some words required my attention, and therefore, I paid attention to them. Basically, I performed a visual scanning process.
With regards to your second option, it didn't call my attention at all since I just saw a block of text with no particular points. If I HAD to read them, maybe I would do it, but without obligation whatsoever, I wouldn't pay any attention at all to that block, because visual scanning was muted or there are not enough stimuli to perform such scan.
Also, in general, it's better to avoid passive voices, and speak as naturally as possible. Your second approach has an instructional vibe that most users will ignore (again, you should test that).
In short: visual scanning transmit content hierarchy. Proper visual stimuli and highlighting of content in blocks of text will help users understand if the content at sight is of value or if they can safely ignore it