@Iceman,
I discussed it a bit in this thread:
That said, I've yet to see any LLMs that really do a good job with seduction material.
They can give basic definitions of stuff, but even when trained tend to still be really off. I have tested some of the better trained ones, like Wingman Live (which a guy on this site seemed to like) and FrenchOG3's, but aside from serving as banter training / helping craft some witty text responses when a guy's not sure what to send, plus giving definitions for things, they are pretty limited.
e.g., if I feed either of them forum posts or even complex text scenarios from Skilled Seducer, I end up shaking my head going, "No, no, no, no!" to the responses.
Basically: if you can imagine a guy who has only ever read a ton of seduction stuff, and memorized a lot of it, and assumes he knows seduction front and back due to all his reading, but has never actually been in-field and does not know what he does not know (and we have had commenters and forum members like that over the years), that is what you get with an LLM.
Dear Friends, Dear Womanizers, Dear Social Artists, Unless you have been living in a cave, you have probably noticed the recent rise in popularity of Large Language Models (LLMs)—so-called “generative AI” (Gen-AI) such as ChatGPT, Gemini, Copilot, and for the more daring, DeepSeek. With the...
www.skilledseducer.com
I would still like to create an LLM that can basically serve as a reference tool. e.g., "I want to know about X; what can you tell me?" and the LLM gives a very basic summary then suggests some articles for reading.
The trouble with that is you need to feed the LLM on articles and program it that "for topic X, recommend these articles", etc. -- LLMs still are not good at figuring that out on their own.
e.g., when I use Brave AI, it will tell me certain things, then link the sources... then when I check the sources, they say something completely different from what Brave AI is telling me. Sometimes I search something on Brave AI twice and get completely opposite, conflicting answers each time (with Brave AI linking the same sources for both!).
So basically, you need a human programmer who will go through a huge list of different things guys may search for and tell the LLM which articles to reference for those. So you need someone familiar with all the articles (nearly 3700 of them) who can then go through thousands of topics and assign them.
It'd be a huge project.
Would be very useful to the site.
But very tangly to put together and not have be a train wreck
Chase
