A (Most Likely Wrong) Theory About False AI Rejections

[OddLove]

I was thinking about how common it's become to use AI generated images to quickly create a avatar/profile picture now, and I had even done that myself before many of our avatars got whited out several months back. But when I went to make a new profile picture I was just bored of trying to get an AI image generator to make the photo I was trying to explain to it, so I went to paint on my computer and just made my own crappy basic one.

Also, full disclosure, I enjoy using AI for silly things, it can be fun hearing Spongebob go hard AF on a UK drill beat, or see Hank Hill kicking Superman's ass.

It's just, personally, I think it's absolutely one hundred percent impossible for One person to review thousands of stories a day, even if they're just quickly skimming through it. So it would make sense that they're using some kind of automated system to filter out tons of stories before humans ever lay their eyes on it themself.

And because of that, my running theory (That I fully admit I have no evidence for, and is most likely wrong) is that a lot of these false positives involving AI rejections could be a system recognizing the avatar/profile picture from the account submitting their story is in fact AI generated.

So to indulge my silly theory, has anyone here who doesn't have an AI generated avatar picture received false AI rejections?

 

[Five_Inch_Heels]

Are we up to 90 threads on this now?

 

[OddLove]

Are we up to 90 threads on this now?

I couldn't find a thread talking about this theory. If you can, post it here.

 

[Five_Inch_Heels]

You couldn't find a thread about machine generated crap? I see at least three, not including this one cluttering up the front page of the AH.

 

[OddLove]

You couldn't find a thread about machine generated crap? I see at least three, not including this one cluttering up the front page of the AH.

Oh, sorry. I misunderstood what you were whining about.

I understand you don't like this conversation, and you feel entitled to have a forum that reflects what you personally want, but I don't think it's reasonable and sane for you to throw a tantrum about it on a thread I posted and derail the conversation I'm interested in having. Nobody is forcing you to click on this. You read the title, knew it had something to do with AI and the mass amounts of false positives happening, and you made a conscious decision to click on it, then you chose to post in it (twice now) which is counterproductive to anyone who truly believes these kinds of conversations are 'cluttering up the front page of AH'.

So if you're offended, frustrated, or annoyed with these topics, you're allowed to start your own conversation topic in your own thread, or even go bump a bunch of other threads to try bumping this one off the front page. But if you're looking for a dopamine boost and complaining helps you get that hit, then you can just respond again and tell me about how this conversation is meaningless or pointless because you deemed it so. I'll read it, and I'll respond to it, and we can see how long we can clutter up the front page, together.

Or just go join conversations you actually enjoy having, it's obviously up to you.

 

[StillStunned]

And because of that, my running theory (That I fully admit I have no evidence for, and is most likely wrong) is that a lot of these false positives involving AI rejections could be a system recognizing the avatar/profile picture from the account submitting their story is in fact AI generated.

So to indulge my silly theory, has anyone here who doesn't have an AI generated avatar picture received false AI rejections?

It's as good a theory as any.

 

[TheLobster]

It's as good a theory as any.

Nah, the others at least purport that the content of the stories themselves matter, which seems like a necessary precondition for a dilter applied to stories.

But sure, let’s speculate in some other directions, too. Who says there’s not an uptick in AI rejections when it’s raining in Boise, Idaho? We may never know!

 

[OddLove]

It's as good a theory as any.

Pretty much what I was thinking, as soon as someone without an AI generated avatar shows up and says they've had false positives, I'm throwing the idea out. I just happened to notice the last few people claiming false positives had AI pictures, so I couldn't not noticed the pattern.

 

[jsmiam]

Interesting question. Definitely a different to other theories.

My prediction is probably not (plus, they wouldn’t know how even if they wanted to) combined with there won’t be enough replies to draw a conclusion.

Next, and not directed at the OP, just me on my soapbox, I’ll raise a point that irks me to no end: I’ve been making humorous images using mostly GIMP but also using whatever free graphics program I had on my computer at any given time the hard way, since the mid nineties. The old fashioned way. Alpha channels. Layers. Dodge and burn. Color curves. Feathering. Despeckling. Anti-aliasing. Whatever.

Now, any manipulated image at all is automatically labeled by everyone as being AI generated. Some of us did it the hard way! And P.S., you kids get off my lawn!

Ok. Thanks. That felt good. Carry on.

 

[Kelliezgirl]

Nah, the others at least purport that the content of the stories themselves matter, which seems like a necessary precondition for a dilter applied to stories.

But sure, let’s speculate in some other directions, too. Who says there’s not an uptick in AI rejections when it’s raining in Boise, Idaho? We may never know!

Fucking Idaho man..

 

[Candy_Kane54]

Not to throw a damper on your theory, but I did get dinged for submitting an AI story and my avatar is not AI generated. However, I do use Grammarly to edit my stories so there must be a check for that going on.

Grammarly has a couple of 'tells' that I know of that you can look for to see if someone is using it. 1. It will always suggest replacing 'nice' with 'lovely' whether it makes sense or not. 2. When it finds a long run-on sentence in the form "Blah blah blah ... , however/therefore/but/since ...blah blah blah," it will always suggest breaking it into two sentences at the 'however/therefore/but/since' with the second sentence starting with "Still, ..." instead of the 'however/therefore/but/since' you used in the run-on sentence.

 

[BobbyBrandt]

And because of that, my running theory (That I fully admit I have no evidence for, and is most likely wrong) is that a lot of these false positives involving AI rejections could be a system recognizing the avatar/profile picture from the account submitting their story is in fact AI generated.

The problem that I have with this theory is that:

1. The forum side of the site and the story side are separate. Hence, an avatar on a forum account would not likely be reviewed when a story is submitted.

2. Profile pictures are moderated and reviewed prior to being approved. If an AI image was used for a profile, it would likely be rejected if the site's policy on AI images was enforced.