The most prestigious law school admissions discussion board in the world.
|
 |
 
|
Pretty impressed with AI's trolling ability
| Lab Diamond Dallas Trump | 02/21/26 |
Poast new message in this thread
Date: February 21st, 2026 7:08 AM Author: Lab Diamond Dallas Trump Obviously it cranks out slop off the shelf, but I trained a LLM using all of my best gaming troll posts over the years and had it learn my overall style and techniques. It can do a deep dive on a forum in seconds and identify the various fault lines and fertile grounds for trolling. Then it proposes the various angles and techniques it will use, I approve or recommend modification, and it bangs out a post in my voice. I explain to it which lines are "AI tells" and make various other edits to improve, explaining my reasoning. It started off doing 3/10 work, but I have it up to an 8/10. What's most impressive is that when I tell it we can't take a certain position because it will cause a hit to my credibility (based on info outside its dataset), the angles it pivots to are clever and things I would use or have used (but that are not necessarily in the dataset).
In a couple of years when things have advanced a lot more, I want to do an XOXO preservation project where you will be able to run a version of "prime" XOXOHTH with AI agents trained on all of the best poaster data cranking out banger thread after banger thread. You'll be able to have it dynamically respond to current events, or relive classic moments like election night 2016. You'll even be able to seed it with whatever forcememes you want to be active and in effect, or have the AI agents create their own forcememes. That will be the free version, but there will also be a frontier version with "Scholarship Mode." |