What percentage of users who have been suspended (for any amount of time) for gpt use were active answerers as you described them prior to chatgpt's introduction?
I wasn't a very heavy flagger of gpt posts, but of the 4 users I flagged... 3 users were previously not active users and the 4th was previously and currently active. The 4th had suddenly started posting well written answers with much quicker turnaround than normal (4 in an hour) that greatly deviated from their normal "Here's some code"-like answers. My interpretation, having seen gpt generated text many times elsewhere (not just here on SO) was that all of these cases were gpt generated and they were all subsequently acted on by mods.
Given I found all of these cases within minutes of opening the bounty tab, I find it hard to believe that these are as rare as your draft-based data seems to suggest. Is it not possible that gpt users simply altered the way they were posting answers? If I open the bounty tab today will I be able to quickly find a few more? (yep, on my first click, new user, 6 gpt answers on bountied questions. no detector necessary.)
The sudden drop is certainly troubling, but I question whether or not giving up is better than continuing to fight against under-verified content flooding the network. Neither alone will win back the users who are leaving the platform in droves. Should we sacrifice quality despite the fact that it won't bring back the users who are leaving, or should we begin to address why they are leaving?