Any Interest in a Zero Local Compute Blog Post AI Text Checker?

in #witness2 months ago


image.png

Came up with a nice neat < 125 lines of code blog post AI text checker that utilizes not 1 but 4 different detection models to estimate if a blog post is AI generated garbage or not.. While not perfect the fact you can run these checks without having to go crazy installing Nvidia AI cards and getting into CUDO / Tensor cores is pretty neat.

An account with more downvote power than I with this thing properly tuned would be a good replacement for human based GPT garbage detection and policing / downvoting such crap..

Anyone want this code released up on github or? It's based off this https://radar-app.vizhub.ai/


VOTE @KLYE FOR WITNESS!

Every Single Vote Helps, Thanks for the Support!

Need to get in Contact with KLYE?
Join the Official #KLYE Discord Server Today!


Looking for an Affordable, Secure & Reliable Server Host for Your Witness Server or Other Web Related Projects? Check out Privex.io!
Sort:  

This is a cool idea, but unfortunately none of the current "AI checker" models actually work in my experience. I just generated a bunch of rants with arty and the highest score any of them got was 0.049 on dolly and 0.2 on vicuna. People will rely on these and they simply do not work. Neat work anyway, its a signal just not a very good one imo

I check against the 4 RADARs available then average it. if it's higher than 98% or so then shit is suspect. It's not foolproof yet sadly.. and to train these RADAR models is time consuming as hell. :/

use advance prompts and it did not work.

I do not believe automating detection of automation will prevent social media from being automated. Looks to me like, absent personal human management, this will eventually result in entirely automated 'social' media that will exclude people entirely.

Thanks!