You are viewing a single comment's thread from:
RE: Making SPAM fighting on Steem into a game of skill
These settings seem like they might be decent for distinguishing between human and whatever they're using... Now if I just had 8 or 10 million SP worth of downvote strength. ;-)
GPT-2(?)
Human
I still don't really know how to read those histograms. It would be nice if it could all be boiled down to a single number, instead of needing to take a SWAG from the coloring.
There was once a time when this kind of downvoting power was supporting the fight against abuse - unfortunately, it got too close to Tron towers and that support ended.
Although this look compelling, it's not the kind of slam dunk you get from plagiarism or spamming ☹️
Agreed. I would love to see an evaluator that produces some sort of level of confidence estimate. As AI gets better and better, this will be an increasing challenge in many areas, not just here. I can imagine it evolving into the sort of "cat and mouse" game that we currently see with malicious software detection.
Now that I know about it, I see a lot of posts and think "that's the kind of meaningless crap that gpt-2 produces".
I've had a bit of a play around with it now and it's really not very good - Although they might be using GPT-3 which I can't find much of for free.
All three of those points match my observations.
Still, GPT-2 is just good enough that there's always a nagging doubt for me, especially as to whether it might be due to a language issue caused by English as a second language or translation with something like Google Translate or DeepL.