GBA Logo horizontal Facebook LinkedIn Email Pinterest Twitter X Instagram YouTube Icon Navigation Search Icon Main Search Icon Video Play Icon Plus Icon Minus Icon Picture icon Hamburger Icon Close Icon Sorted

Community and Q&A

Are the chat-bots using us for practice?

DCcontrarian | Posted in General Questions on

I feel like a lot of the comments lately have the hallmarks of AI-generated text.

Typically they include: 
* Comments on a thread that’s been dormant for years, by a user with no posting history. 
* A stereotypical four-paragraph style, where the first paragraph is some variant of “this is an interesting subject” and the final paragraph is some variant of “of course you should consult a real professional.”
* Advice that is not very specific, but also not very correct. Often it doesn’t address the question asked, it just offers some platitudes on the subject of the question.

A subset of the responses are the ones that are spamming a product or service, they typically have a link in paragraph 3.

Anyone else noticing this? 

It seems that the best defense would be to lock threads once they’ve been dormant for a certain amount of time.

GBA Prime

Join the leading community of building science experts

Become a GBA Prime member and get instant access to the latest developments in green building, research, and reports from the field.

Replies

  1. Malcolm_Taylor | | #1

    DC,

    Yes - and this is early days. I doubt things will improve.

  2. nickdefabrizio | | #2

    This is one reason why I always use my full name in any chats I am involved with. If I don't feel comfortable with people atttributing something to me, I generally should not be saying it in public. I always advise my clients that they should assume before they post anything or email anything that the post/email will be read by their wife, mother, grandkids....and a judge..

    Wait until full articles in some lesser quality "home improvement " magazines are written by these programs-if they are not already..

    1. KeithH | | #4

      It would be interesting to see GBA have a verified check and user flairs like DIY, homeowner, pro, (lots of pro sub tags should be possible)

      Personally I am not a pro and prefer a small amount of anonymization.

      1. nickdefabrizio | | #10

        That could be interesting. I am always amazed at how broad the expertise is on these forums and would be curious to know about the professional backgrounds of particular responders. On the other hand, for me they would have to create a user flair like: "gadda-bout"....or "absent minded DIY"......

  3. KeithH | | #3

    This is an interesting question. I disagree with the idea of locking dormant threads. Sometimes users come back to update very old threads with information about completing a build etc. Of course, you should see what the GBA editors say.

    But all formatting jokes aside, I’ve received answers/updates to old questions multiple times that were pertinent and informative. I’d rather see old threads stay open.

  4. DennisWood | | #5

    I’d vote for not locking threads. I’d also agree that a lot of content I see these days is not written by a person.

  5. gusfhb | | #6

    Actually locking thread is a good idea.
    If you have an update, start a new thread, with a link in your first post to the old one.
    Newbies tend to dig up old threads and it gets silly
    I participate in many forums and zombie threads are an issue in all of them

  6. Expert Member
    DCcontrarian | | #7

    One other characteristic -- they don't actually participate in the conversation. No give and take.

  7. charlie_sullivan | | #8

    I like the idea of locking threads after perhaps 3 months. The original poster can post a new thread and link to the old one if a followup is needed.

    When somebody new adds a follow-up question to and old post, I often see the original question and start reading responses without realizing that it's an old thread.

  8. kbentley57 | | #9

    I'd actually say I was impressed with how well some of them do. It's pretty obvious to us when it's a bot, but on the other hand, we're not getting garbage text with links to fake rolexes any more!

    I think a good system would be a 'report post' functionality with N number of reports deleting the post.

  9. maine_tyler | | #11

    " It's pretty obvious to us when it's a bot"

    I guess I'm in the minority where I gotta honestly say I don't know what posts people are referring to. Like lots of them on recent threads? Or just a peppering across months worth of threads?

    Sometimes human responders follow similar patterns outlined by DC. haha.

    1. Expert Member
      BILL WICHERS | | #12

      A typical problem post would read something like "great discussion, thanks for that info. I had really good service from super-building-fixers-that-spam-on-GBA.com, and I'd give them a call to check on your project before you get too far into things". Etc.

      Generally if there is a link to some kind of contractor or consultant, it's probably spam. If the post is years later than the rest of the discussion, and isn't one of the original posters with an update, chances are it's spam UNLESS it's a question about whatever the original topic was. The spammers are usually just trying to squeeze a link to their product or service into the Q+A thread, so the text of the post is just babble somewhat related to the thread, but the object of the post will be to get the link in there for marketing purposes. If you read a few of these, you'll soon spot the pattern they follow.

      Bill

      1. Expert Member
        DCcontrarian | | #13

        Keep in mind that the moderators edit subject headings to try to get as many keywords that people might search for, so that the pages are more likely to hits in search engines. This makes the inline advertising more valuable, which is what pays the bills. What the spammers are doing is trying to sneak in free advertising. And they're taking advantage of that keyword management.

      2. maine_tyler | | #15

        Sure, but maybe what I'm saying is it doesn't feel like the ratio of bot/spam posts to real posts is that high. Perhaps I'm just not paying enough attention, and perhaps it's getting worse...

        1. Expert Member
          BILL WICHERS | | #16

          I think the concern is that it's picked up a fair bit over the past several months, which is a bit of a cause for alarm. I don't know how many on here remember Usenet from back in the day, but spam there started as a relatively few posts, but ultimately got so bad that it basically killed Usenet as a useful communications medium in many cases. We don't want to see that happen here.

          Bill

          1. Expert Member
            DCcontrarian | | #17

            I'm also old enough to remember Usenet and you've hit on exactly what I'm worried about. Online communities can be fragile, they can be easily destroyed.

  10. dan_saa | | #14

    I vote for not locking threads. For example I've followed up on my own post with new info that I've learned a couple years after original post. Starting a new post seems like too much work to set the context for the updated information that I'll be less inclined to post an update.
    (domain root greenbuildingadvisor ) /question/epoxy-vapor-retarder-over-e-concrete

    The cadtutor dot net forum has a user reputation system where you can heart/start/award posts. If something like that was in place, then new user posts with minimal reputation could be treated differently, perhaps not allowing adding links.

  11. andyfrog | | #18

    I vote for not locking threads, for all the reasons outlined above.

    I also feel that if we do lock old threads, the bots will just start new spammy threads instead, and if I had to choose, I'd rather read a thread list of old threads bumped by bots with useful info, than a whole thread list of spam threads.

  12. Expert Member
    DCcontrarian | | #19

    Post #25 in this thread -- https://www.greenbuildingadvisor.com/question/please-check-my-plumbing-work -- is an example of what I'm talking about:

    "I suggest you take a look at other similar projects and see how they are done. This can give you an idea of what works and what doesn't."

    Two months after the last posting on the thread. Not in response to anything in particular. Pretty much reads like what you'd get if you posted the original question to ChatGPT.

  13. Expert Member
    DCcontrarian | | #20

    That you, ChatGPT?

  14. AndyCD | | #21

    OMG. Perfectly illustrates your original contention, except ol' chattie got it done in one paragraph instead of four. What irks me is that user "Chromosedry" is listed as a member. Does that mean a paid subscriber?

Log in or create an account to post an answer.

Community

Recent Questions and Replies

  • |
  • |
  • |
  • |