Armed Polite Society
Main Forums => The Roundtable => Topic started by: lee n. field on April 17, 2024, 12:03:14 PM
-
(Copy pasted from my BookFace.)
How much of what we see now in the internet is AI generated jabberwocky?
Was looking around at what it takes to get the Thunderbird mail client to talk to Protonmail*. (I'm old school, and think email should ideally be done through a local program instead of a remotely hosted web interface.)
I find this, purporting to be a comparison between the two. It's a category error. These two things you wouldn't compare. It gives the appearance, sort of, of being helpful, but really isn't. The reviews look sketchy. "David S. Verified user in Computer Hardware" -- really?
https://www.g2.com/compare/protonmail-vs-thunderbird (https://www.g2.com/compare/protonmail-vs-thunderbird)
I suspect it all AI generated.
How much of what we see is? This is far from the only thing I've seen like this.
(*Turns out, they have a "IMAP bridge" program to make this work. Which, if it works, I can live with. So, I'll probably end up going with the pay version of Protonmail. Just, looking forward past my retirement date.)
-
More and more.
Up until now, I have been pretty good at spotting the AI stuff, just because once it gets to be more than a sentence, it starts to become "off". However that is getting more and more difficult to spot.
One example is that Google's Gemini actually sounds like a real live commie pinko google wonk when I call it out on bias. Hard for me to tell the difference between what it is arguing and what a smelly, Birkenstock wearing hippie would argue.
Another example is I note on Amazon reviews, at the top of the reviews page, there is now a preface blurb giving an executive summary of what all the reviews are saying. It states that it is AI generated, but it's pretty difficult for me to tell, and most of the time I wouldn't be able to, were it not for the AI disclaimer.
Scary times.
EDIT: Also, I posted a while back about a SHTF series I was reading, where the last book of the series went from a four star average to a two star, with the major complaint being that it was a completely different writing style from the previous books, and would go into agonizing detail about stuff like the color of a wall. Most reviewers were betting the author called the last book in via AI. I have also recently seen several fiction series on Amazon where the authors are pumping out like twenty books a year. That just seems kind of high for a human author to do without at least AI assistance (like the AI pumps out the book in five minutes, and the author spends a week making minor changes to make it sound not so AI).
-
All of it.
Including this post.
-
A lot of the sponsored stuff on FB will have a hundred comments, all glowing, all one-liners.
I suspect that a lot of the guys who are cranking out 20 books a year are doing it as a full time, job... They also tend to be writing to a mold/pattern... Crank out an outline, you already have the characters mostly scripted, then fill in the outline 5,000 words at a whack...
Mickey Spillane would write when he was running low on money. I think he cranked out one of is pulp novels in about two weeks. Before word processing.
-
All of it.
Including this post.
I knew it! Cordex is AI everyone!
-
The overlap between the smarter AI and the dumber humans is getting larger every day.
It's similar to the problem with bear proof garbage cans.
-
I knew it! Cordex is AI everyone!
Beep boop.
-
I’ve noticed many news articles now read the first half like a stupid online recipe where the author (or AI author) jerks off for a few hours pages about idiotic bullshit for word count