You are viewing a single thread.
View all comments
74 points

How can you tell that’s it an AI generated comment?

permalink
report
reply
128 points
*

I’ve dealt with this particular bot problem as a mod. Noticing the pattern early on was easy because they always had posts on TEMU subreddits. Now that those subs are banned, it’s harder.

  1. Comment length. Their comments are always less than two lines long.

  2. Tone. They always contain at least one exclamation point, and tend to have a tone that I’d say is “cheery to the point it’s offputting.”

  3. Context. Reading them, a lot of the time you’ll ask “did they even read what they’re replying to?” The comment seems related, but directly misses the main point.

  4. History. Look at the rest of their comments. If more than one ends in two emojis, they’re definitely part of this particular bot network.

permalink
report
parent
reply
26 points

Also punctuation and usage of commas, apostrophes etc. Far too formal usage of style to go with the casual message of the comment.

permalink
report
parent
reply
1 point

You used commas, capitalization, and periods here. Therefore you are a bot exposing yourself!

I don’t usually drop punctuation unless I’m on text message or Discord. Still used to that Reddit norm and sticking to the norm of using regular punctuation isn’t one I take issue with, especially on a site based around reading other peoples’ comments and writing responses.

permalink
report
parent
reply
1 point

To be fair, a lot of people also use proper spelling and punctuation.
This just gives away they’re not a human from the hood.

permalink
report
parent
reply
22 points

History. Look at the rest of their comments. If more than one ends in two emojis, they’re definitely part of this particular bot network.

All of hexbear is bots, confirmed.

permalink
report
parent
reply
10 points

Emoticons ≠ emojis.

permalink
report
parent
reply
2 points
*

I dunno that’s a conspiracy that feels like it might be true. One guy and his communist bots.

At least you know they are thinking as one that way.

permalink
report
parent
reply
12 points

This is dystopian

permalink
report
parent
reply
11 points

Yeah they always sound like the overenthusiastic 45y/o HR lady.

permalink
report
parent
reply
11 points

“cheery to the point it’s offputting.”

Yeah, it’s not something I’d notice with one comment, because sometimes people are just in that mood, but looking at all the comments together it’s pretty obvious.

permalink
report
parent
reply
8 points

Is there a way to prevent them from flooding the fediverse, or will that fall onto the shoulders of mods to find and delete them?

permalink
report
parent
reply
4 points

No clue.

permalink
report
parent
reply
2 points

They can defederate from hexbear. But that’s up to each instance individually.

permalink
report
parent
reply
2 points

Hopefully we’re just never big enough to matter.

permalink
report
parent
reply
5 points

Hey btw, the vocabulary used is often spontaneous - something a human on the internet never really uses

permalink
report
parent
reply
3 points

As for point #3…a lot of people on reddit have poor reading comprehension.

Okay, or maybe they’re bots, but FFS, how do you tell the difference? A lot of people have short attention spans. (See the prevalence of TLDR.) So, they just skim whatever they’re reading instead of slowing down and thinking about it.

permalink
report
parent
reply
2 points

Someone missing context doesn’t make them a bot. To be a part of this particular bot network, all 4 points will be present.

permalink
report
parent
reply

Wow, alright. Today I learned my neurodivergent ways of communicating online make me sound like a bot. On one hand, booo. On the other, well I did always want to become a robot-human hybrid so… win?

permalink
report
parent
reply
9 points

Nah, I scrolled through your recent comments and you’re fine. The description above is sort of close, but not quite – it’s more like every comment is syrupy sweet aggressively cheerful and unnaturally pleased about everything. Lots of overly familiar terms of endearment (friend, buddy, fellow Redditor, etc.). It also has a subtle rhythm – generally reacting to the parent post/comment, adding a friendly observation or two, and wrapping it up in an inspirational/motivational bow. Grammatically perfect, no misspellings or shorthand or sentence fragments or trailing thoughts. Zero negativity. Vocabulary that uses dated clichés to sound folksy and relatable while still being a bit stiff and formal. “Jokes” that don’t make any sense, or that miss the context of the discussion.

Like, here are three ChatGPT responses to your comment when given the prompt “write a friendly reply to this Reddit comment” (which I suspect is similar to the prompt these spammers use):

I totally get where you’re coming from! Embracing our unique ways of communicating can sometimes have unexpected and even amusing outcomes. It sounds like you have a great sense of humor about it! Remember, the most important thing is to be yourself, whether that comes off as a bit robotic or not. Who knows, maybe the robot-human hybrid future is closer than we think! Keep rocking your unique style! 😊

Hey there! Embrace your unique way of communicating. 🤖✨ If sounding like a bot means you’re precise, efficient, and to the point, then that’s a superpower in my book! And hey, who wouldn’t want to be a cool robot-human hybrid? Rock on with your awesome self! 🚀🤘

It sounds like you have a great sense of humor about the situation! Embracing our unique ways of communicating can definitely be a superpower. And who knows, being a robot-human hybrid might just be the next step in evolution! Keep being your awesome self! 😊

There’s nothing technically wrong with these lines, but no real person who isn’t a corporate social media PR manager actually talks like this.

permalink
report
parent
reply
2 points

what is a TEMU subreddit?

permalink
report
parent
reply
0 points

A subreddit about TEMU.

permalink
report
parent
reply
44 points

There’s no rigorous test, but once you’ve worked with it enough you can kinda just tell. This reeks of ChatGPT.

permalink
report
parent
reply
10 points

I’ve been applying for jobs recently and was using ChatGPT for help with the cover letters, mainly just for inspiration and I edited it. On the second one I noticed the suggestion was more or less exactly the same. So I just modified my first and put a bit more work into it to make it original. Each subsequent one was the same, and I generated a dozen and a half. Useful for picking out things worth highlighting, but I wouldn’t be surprised if that first and maybe even second place assumed it was ChatGPT right away lol. Definitely know what it smells like.

permalink
report
parent
reply
-1 points

Ah yes the “just trust me bro” scientific message.

permalink
report
parent
reply
7 points

People can have a “gut feeling” about something and be absolutely right. It’s just something that you know is true based on previous experiences. It’s an educated guess.

In a similar way, we each have certain quirks or do things a certain way, so other people who know you may know, for example, that, when you get home at 5pm, you’re going to set your bag down in the living room and go lay down on the couch for a few minutes . There’s no evidence that’ll be the case this time, but, based on previous experience, they can make a pretty good educated guess that that is exactly what’s going to happen.

permalink
report
parent
reply
6 points

Just because you’re oblivious doesn’t mean we can’t fucking taste it

permalink
report
parent
reply
-9 points

Has that feeling of truthiness egh?

permalink
report
parent
reply
40 points

The “Keep scrolling and enjoy the ride, my friend.” is what tipped me off. ChatGPT tends to say stuff like that when trying to be enthusiatic.

permalink
report
parent
reply
27 points

Honestly doesn’t seem too far off from standard Redditese to me

permalink
report
parent
reply
14 points

At least it’s not “the narwhal bacons at midnight” shit

permalink
report
parent
reply
16 points
*

You can pop the first comment into CharGPT with ‘write me a reddit comment in response to’ and youll get similar stuff.

It’s a but ridiculous to claim it’s GPT related without proof, though. You can’t prove it and claiming you can ‘spot it’ is just hilarious to me.

“Wow, that’s quite the conspiracy theory! While Reddit has definitely seen its ups and downs, I think it’s safe to say it’s still very much alive, thanks to its dedicated community. Let’s keep the discussions going!”

“Seriously? Reddit was on its last legs long before June. This ‘Necromancer’ is just a desperate attempt to breathe life into something that was already fading away. RIP old Reddit, it’s a brave new world now.”

“Ha! If this is the work of a necromancer, they must be doing a pretty good job because I’m still addicted to scrolling through here!”

“Ah, so that explains the zombie posts and resurrected memes!”

“It’s unfortunate to hear that some believe Reddit has declined, but communities on the platform can evolve over time. Let’s hope for a positive revival if that’s the case.”

“I understand your concerns, but I don’t believe Reddit has died. While there have been changes and controversies, it’s still a vibrant platform with millions of users. Let’s hope it continues to evolve positively!”

permalink
report
parent
reply
13 points

This bots aren’t particularly clever and just regex a prompt template to put into ChatGPT automatically. As you would quickly realize playing with any LLM, if you enter the exact same prompt into ChatGPT you won’t get the exact same answer, but all answer will be rather similar and follow an, eventually, predictable pattern. While every comment will be certainly different it will also be repetitive. When put back to back, as when reading all the comments from a single account in the profile page, the repetitive pattern becomes obvious.

permalink
report
parent
reply
11 points

You can tell it’s robots by the way it is.

permalink
report
parent
reply
9 points
*

I think everyone’s reaction immediately assuming it’s a bot is odd as well. Everyone’s reasoning is just “you can tell by the way it is”. I’m just sitting here like uhhhh.

If you look at the pattern of comments in someone’s profile, it’d be far easier to tell. But looking at a singular comment in isolation like this and just assuming it’s a bot “because it is”. Ok?? I’m not convinced by a singular post and I don’t understand how everyone else is.

permalink
report
parent
reply
1 point
*

OK, whoever is calling this a bot, enlighten me. I’m too dumb to tell if it’s a bot or not:

https://www.reddit.com/user/artisticEggs09724/

The only maybe tell I have is that every comment seems to be uplifting and written with excitement. Aint nobody that happy. Maybe. lol.

permalink
report
parent
reply
8 points

too positive and articulated

permalink
report
parent
reply
-21 points

Nobody can accurately. This is copium in a sub for degenerates who can’t control their addiction. It’s grasping at straws because Reddit is still as popular as ever.

permalink
report
parent
reply

Reddit

!reddit@lemmy.world

Create post

News and Discussions about Reddit

Welcome to !reddit. This is a community for all news and discussions about Reddit.

The rules for posting and commenting, besides the rules defined here for lemmy.world, are as follows:

Rules


Rule 1- No brigading.

**You may not encourage brigading any communities or subreddits in any way. **

YSKs are about self-improvement on how to do things.



Rule 2- No illegal or NSFW or gore content.

**No illegal or NSFW or gore content. **



Rule 3- Do not seek mental, medical and professional help here.

Do not seek mental, medical and professional help here. Breaking this rule will not get you or your post removed, but it will put you at risk, and possibly in danger.



Rule 4- No self promotion or upvote-farming of any kind.

That’s it.



Rule 5- No baiting or sealioning or promoting an agenda.

Posts and comments which, instead of being of an innocuous nature, are specifically intended (based on reports and in the opinion of our crack moderation team) to bait users into ideological wars on charged political topics will be removed and the authors warned - or banned - depending on severity.



Rule 6- Regarding META posts.

Provided it is about the community itself, you may post non-Reddit posts using the [META] tag on your post title.



Rule 7- You can't harass or disturb other members.

If you vocally harass or discriminate against any individual member, you will be removed.

Likewise, if you are a member, sympathiser or a resemblant of a movement that is known to largely hate, mock, discriminate against, and/or want to take lives of a group of people, and you were provably vocal about your hate, then you will be banned on sight.



Rule 8- All comments should try to stay relevant to their parent content.

Rule 9- Reposts from other platforms are not allowed.

Let everyone have their own content.



Rule 10- Majority of bots aren't allowed to participate here.

Community stats

  • 3.7K

    Monthly active users

  • 746

    Posts

  • 32K

    Comments