Subscribe
Sign in
How to detect bots' comments in your PH discussion? π€
I am here for answers. π
Daily, my conversations are flooded by fake profiles. (I am pretty curious how many I will be having under this post.)
There are still some signs that detect them, but it is time-consuming and not so easy.
This could help with detection:
π Generic profile photo (saw profile photos of some celebrities).
π That bot repeats keywords from your original post.
π Non-sensical names.
π They miss parts in the text (e.g. Our page [insert link]... :D)
π Their profiles are without information.
π If they have social media, they are empty as well.
π Their Activity tab in profile: You can see many comments written at the same time (suspicious and impossible).
What else?
45
Replies
Best
Aruna Neervannan
Run the reply through chatGPT and ask it to identify? :)
Report
Share
I was feeling the same thing. I think if you used LLMs enough you recognize their tone
I wanted to add this point too, but then I realized that sometimes I use gpt to check my English grammar, and my message could sound like I'm a bot, but I'm not π
π
@busmark_w_nika I asked for feedback on my project and i got this comment "Your project is like a finely crafted code - ready to launch and make waves in the digital world! Time to let it spread its tech wings and soar!" I am 99% sure it's generated by an LLM. for me when i am seeking human interaction i want to interact with other humans, share our thoughts ... comments like these are just meaningless
@kksrm I think that's alright. I would prompt it to keep my voice and fix the grammar
@hafid0x00 Yeah I wonder what was the person that gave this feedback thinking? Would he/she want to receive such meaningless "feedback" for their project...?
@xspiral @busmark_w_nika I don't think so Nika. Wasn't pointing at you at all! π
Just saying there's a lot of AI generated, superficial content on PH posted by real humans.
I find it intriguing how GPT bots, with their remarkable ability to generate text, have become the virtual chameleons of conversation. Certainly, it's a marvel how they weave words into sentences that are both perplexing and delightful. What's more interesting is their knack for turning even the most mundane queries into epic narratives, replete with vocabulary that sounds like it was plucked straight from a thesaurus on a caffeine high. It's almost as if these bots attended a prestigious language academy in a parallel universe where the motto was, "Why say 'good' when you can say 'exceedingly superb'?" π
@busmark_w_nika Which one? Bye bye bye? or was that n*sync π€ (also please don't report my comment, I admit it, it was generated by a bot, I am guilty as charged π)
@busmark_w_nika I think I need to borrow a proper Sony discman w/ 10sec anti-skip-memory and AAA batteries. To really get the authentic experience and learn what you guys went through ^^
There are the whole networks of them
Easy to spot if you check the Activity tab in their profile - they'll have the exact same product upvoted in the exact same order
Hey, Nika! Really good points, thank you!
And what do you think? Do we need to report it if we're really sure that it is a bot?
Once, I also had a comment that wasn't relevant to the discussion topic and asked me to open a strange link. I decided to report it. Do you think that was a good decision?
Yeah I can definitely spot AI constructed comments too. Just by looking at the words they use. Totally over the top stuff like "your 1K kitty coins are like the digital equivalent of striking gold in a cat-filled mine!" πππ
I'm curious why anyone would create these bots to go around accounts and post comments in random discussions. Are they really trying to build activity in this way to use them for supporting unfair projects in the future?
@busmark_w_nika Did someone create a profile that looks like it's real?
There is a great article by Krissy Davis on WeAreDevelopers "How to Tell if Something Was Written by ChatGPT"
The main points for me:
- Keywords like: βcutting-edge softwareβ or "ever-evolving landscape"
- Unusually formal tone in text that's meant to be conversational or casual
- A remarkable absence of meaning in the comment π§
Thank you for this checklist, was looking forward for something like this. Bots are starting to be real pain in the ass. Sometimes I even wonder if it's the beginning of end of internet :(
EDIT: I use AI heavily for myself for ideation and research but flooding forums with shitty comments is not the use of it I would ever approve
This comment highlights the practical benefit of detecting comments from bots and uses an action verb hook to grab attention. It also mentions the target audience visitors to showcase relevance.
"When you know β you know"
Let's see how many bots are going to post comments on this discussion... π€£
I know what you mean, those bot comments are so annoying! π Another thing I've noticed is a lot of them have gibberish or randomly generated usernames. And if you click on their profiles, they were often created super recently and have barely any info. Definitely suspicious! π€ I wish there was an easy way to auto-detect and filter them out. Maybe some kind of AI could help identify the patterns? π€ Let me know if you find any good solutions!
I also realized three things
1. Their replies are long/ verbose.
2. They'll summarize your post (e.g. "Great question Nika! Detecting bot comments is a real challenge especially as LLM technology proliferates bla bla bla")
3. The advice is generic (whereas you can tell when the advice is real because it contains some anecdote)
