zlacker

[parent] [thread] 2 comments
1. fragme+(OP)[view] [source] 2023-08-24 18:42:18
Given the advancements of LLMS, have you given thought to automating some moderation to tell the user they're about to leave a predictable repetitive flamewar comment? Ie, a cleverer version of https://blog.xkcd.com/2008/01/14/robot9000-and-xkcd-signal-a...
replies(2): >>dang+d1 >>sudden+WW
2. dang+d1[view] [source] 2023-08-24 18:49:13
>>fragme+(OP)
Not yet, but the relevant data for doing this is mostly public, and if anyone wanted to work on it, we'd certainly be interested in what they came up with.
3. sudden+WW[view] [source] 2023-08-25 01:02:10
>>fragme+(OP)
Interesting solution. I had some concerns at first but increasing the threshold and manually whitelisting some common valuable but non-unique comments would probably go a long way. Maybe even reduce it to parts of a comment to reduce lazy jokes.
[go to top]