Page 1 of 1

Horrified with AI deciding for us.

Posted: Mon Jul 22, 2024 5:54 pm
by Rad_Schuhart
Hi all. Long story short, the software in my website has updated, and there are some new windows and things that I have not seen till now.

I saw a new menu saying that I got more than 10.200 spam messages, and I could check only the last 3 of them. I opened them up, and they were from a friend, commenting my entries.

I panicked for a second and wondered how many messages more I got deleted by those AI spam filters. I tried to disable those filters, that claim to have a 99.99% of accuracy, and to my horror, I can't. It is not like I don't know the how to, but I am not allowed, because somebody, somewhere, believes that an AI is more inteligent than us, mere mortals.

At least I have seen that, unless the AI considers that it is an horrific spam with terrible content, I can check the spam I get for 2-3 days before it gets deleted. But again, that leads to the main issue, what if what the IA believes is horrible content, is a message of a friend? Yesterday I checked that folder again, and I saw a message in the spam folder. It was from Ian, a friend from this forum, that left me a wonderful comment on my stencils page. Comment that was filtered and in the process of being deleted if it wasnt because I saw it last minute.

Yet again, I wonder how many of those 10.200 deleted coments were spam or messages from real people.

If you left me a message and it never showed up, now we know why. Sorry :(

Weird times we are living...

Re: Horrified with AI deciding for us.

Posted: Mon Jul 22, 2024 9:50 pm
by tankme
It's weird because I get spam messages from spoofed contacts somehow. If I look at my spam folder, I'll see messages from an old co-worker and like Dean Goodin from DAK. Issue is, the messages aren't from them. The contact name is the only valid part of the Email. Everything else is spoofed. Also, my Yahoo account will move stuff to my SPAM folder if I don't read Emails from the person. So I get some ads that occasionally I like to look at the specials. If I move too many of them to the trash without reading, it marks them as SPAM. So now I'm constantly looking to make sure stuff isn't going there that shouldn't. This era of AI is going to be interesting to say the least.

Re: Horrified with AI deciding for us.

Posted: Tue Jul 23, 2024 2:33 am
by Herr Dr. Professor
AI = Annoying Interruptions

Re: Horrified with AI deciding for us.

Posted: Tue Jul 23, 2024 6:01 am
by Kaczor
AI = Anonymous Indian

Re: Horrified with AI deciding for us.

Posted: Tue Jul 23, 2024 7:07 am
by Rad_Schuhart
Spoofed weird messages is something I got in my email, but in this case it is real true people that wrote me.

I wonder why the AI believes this is spam.
Captura de pantalla 2024-07-23 080752.png
Captura de pantalla 2024-07-23 080752.png (38.93 KiB) Viewed 750 times
Those three messages were marked as spam and going to be deleted.

Re: Horrified with AI deciding for us.

Posted: Tue Jul 23, 2024 1:20 pm
by RobW
They're short sentences, with very few word links to the website coupled with multiple posts in a fairly short period. Unfortunately that's what you'd expect from a positive feedback comment.
Ironically the AI generated spam would have scraped the topic and written something that would pass the checks.... The posts I see on a work related forum are very impressive. There's usually something odd in the syntax to give it away, along with links hidden in full stops for a variety of sites we don't want to on our pages.

Re: Horrified with AI deciding for us.

Posted: Tue Jul 23, 2024 2:13 pm
by ColemanCollector
I'm getting lots of celebrity photos lately. I had no idea so many of them had 11 fingers, club feet, and 1m long forearms!

Mike.

Re: Horrified with AI deciding for us.

Posted: Wed Jul 24, 2024 6:59 am
by Rad_Schuhart
RobW wrote: Tue Jul 23, 2024 1:20 pm They're short sentences, with very few word links to the website coupled with multiple posts in a fairly short period. Unfortunately that's what you'd expect from a positive feedback comment.
Ironically the AI generated spam would have scraped the topic and written something that would pass the checks.... The posts I see on a work related forum are very impressive. There's usually something odd in the syntax to give it away, along with links hidden in full stops for a variety of sites we don't want to on our pages.
The thing is, those friend's messages are very different from the real spam ones I get and get filtered as well. Those spam messages and comments share the same several things in common, like a lot of numbers (you won 30.000.000 brazilion Euros!) numbers of pills, and, sigh, grams. And all of them have also at least one long link. It is completely different to the messages I posted above.
ColemanCollector wrote: Tue Jul 23, 2024 2:13 pm I'm getting lots of celebrity photos lately. I had no idea so many of them had 11 fingers, club feet, and 1m long forearms!

Mike.
Well yeah, the IA pictures are another story, lol. But at some point we won't be able to know what is real and what is not.

Re: Horrified with AI deciding for us.

Posted: Wed Jul 24, 2024 9:22 am
by RobW
On your site, yes. I'm seeing spam posts on the work external forum which are trying to score points to elevate a user's profile. So, "nice model", "excellent answer" and the like. Most of the tools are generic and/or trained on a wider selection of answers.
If you put a survey/post option into any receipts you send out can you use those to bypass the filters?