AI Bots are coming. What is the plan?

D

dvduval

Active member
Let's face it. There will be fake users joining our forums in the near future to do all kinds of things, surely more than I can imagine:
  1. Harvest content
  2. Make friends with our users, and scam them
  3. Find ways to post backlinks or promote something
  4. maybe even sabotage our communities in some way.
Further, the bots are already scavenging our sites for content for the their LLMs, so basically stealing our content for future use.

So what is to be done...

Some of my ideas:
  1. Require login for a user to visit more than X content pages (maybe 10)
  2. Cloudflare and any other tools that will help to identify and stop bot behavior
  3. Potentially closing off some portion of the site to the public (just how to get ad revenue on non-public content). One strategy would be to identify the pages that are rarely visited and require login to see those pages.
Is anybody even thinking about this?
 
If your content is public then anyone can digest it, learn from it and use it. AI is really no different in that respect so I wouldn't worry about it. AI created content is obviously less desirable unless it's declared as such.

Placing restrictions on guest viewing is irritating and may deter someone from registering. Having and advertising a private section for members is far less detrimental and may even encourage registration plus it stops the AI bots scraping that area.

Personally I'd like to see forum platforms incorporate AI for better functionality because the potential is limitless. Hopefully the focus will move away from content creation and more towards providing tools to improve on content delivery rather than creation.

Old Nick said:
The question is how to recognize them... :rolleyes:
Click to expand...
At this point in time it's really not that hard. If your forum has a good sense of community and proper interaction between members, AI generated content is reasonably easy to spot, maybe not as a one off first post but if you interact with it, it soon becomes apparent.
 
zappaDPJ said:
At this point in time it's really not that hard.
Click to expand...
Currently it is quite easy, indeed in a few messages we can suspect an AI, but this will be more and more difficult as technology progresses, and probably we will reach a point where only an AI will be able to recognize an AI.
 
Last edited:
zappaDPJ said:
If your content is public then anyone can digest it, learn from it and use it. AI is really no different in that respect so I wouldn't worry about it.
Click to expand...
There is a huge difference: If someone individually learns from my content that is the intention of the forum. But AI (and the companies behind it) are commercially exploiting my content: They make money from it (that's part of their business model), they reproduce it at scale on their platforms and models and this way my content is misused and people have no necessity to visit the forum and possibly participate in it - they just ask the AI and get delivered whatever they are asking for. The AI companies make money while my forum members and me, who created the information, sometimes in years of work and research not only get nothing out of it but get - even worse - punished by loosing visitors.

In my opinion AI companies are parasites for forums and as such should be locked out where ever possible.

dvduval said:
Further, the bots are already scavenging our sites for content for the their LLMs, so basically stealing our content for future use.
Click to expand...

zappaDPJ said:
Placing restrictions on guest viewing is irritating and may deter someone from registering. Having and advertising a private section for members is far less detrimental and may even encourage registration plus it stops the AI bots scraping that area.
Click to expand...

The latter is what I am actually doing. It fosters registrations but also fosters the number of users that do only leech and not participate actively (post). The other downside is, if the higher quality content of my forum is moved behind the registration barrier it will contradict the idea to be found by users via search engines as those are locked out from indexing the content, too. But possibly that is the price I have to pay.

Other than that I use the "known bots" plugin, monitor the bots visiting my forum through it and adjust my robots.txt regularly. Threre are however a number of bots that simply ignore the robots.txt, so possibly I have to lock those out via .htaccess. And obviously you have always the time and effort topic of maintaining those plus you won't recognize all bots.
 
I see a ton of content in my vertical that is clearly harvested by chatGPT. It seems we have reached the point where just rewriting things means they are not plagiarized. What I think a bot can never easily do is replace real people, so doing more to humanize our communities and help people to feel like they have real friends is key. They will surely value the opinion of humans (in many cases) more than bots.

But still I know a lot of bandwidth is being consumed by these bots, and I know that bots that try to emulate humans posting on my forum is inevitable, so just trying to think a little ahead for what is coming.
 
dvduval said:
Let's face it. There will be fake users joining our forums in the near future to do all kinds of things, surely more than I can imagine:
  1. Harvest content
  2. Make friends with our users, and scam them
  3. Find ways to post backlinks or promote something
  4. maybe even sabotage our communities in some way.
Further, the bots are already scavenging our sites for content for the their LLMs, so basically stealing our content for future use.

So what is to be done...

Some of my ideas:
  1. Require login for a user to visit more than X content pages (maybe 10)
  2. Cloudflare and any other tools that will help to identify and stop bot behavior
  3. Potentially closing off some portion of the site to the public (just how to get ad revenue on non-public content). One strategy would be to identify the pages that are rarely visited and require login to see those pages.
Is anybody even thinking about this?
Click to expand...
You will have to harness AI solutions to fight the other AI... It is going to get interesting.
 
They have been here for quite a while already.
Attempts at classified ads fraud are WAY up on both of my forums. I have a tool in place to usually identify them in seconds, from XF2addons. (Direct Message Phishing Detect) They always seem to email 5+ people within minutes, asking for money to be sent to a payment by email service, for the item in the listing. (from recent ads where people are looking for items, etc) They could even be real people, I don't know. I turned off content being emailed when someone does a new DM, but just the title and a link to come back. By the time they get the email, I have usually intercepted a scammer and spam cleaned all of their content. If it was emailing the content, or the staff wasn't quick to respond, there could be some real damage. Disclaimers galore are posted, with tips to spot fraud, for our members.
We also get accounts set up through automation that reply to posts from years ago with a reply that sounds very AI like, with no real point or question.
Maybe a new add-on that can detect AI generated content would be a step in the right direction? Of course, we have an actual chatgpt subforum, where people can ask questions, so there would have to be exceptions for where it does its detection :)
 
The forum that I started has turned into more of a digest than an active forum, unfortunately. To bring it back from its death throws, I have utilized the AI, even going so far as to sort all content and provide answers and mark for best solutions to long standing threads, but it took hundreds of hours to get to the stage its at, and so far it has yielded no results. I believe strongly in original content and more people, so its no substitute for that. I am hoping an abundance of regular AI-generated content interests people, as well as the ability to get answers to common issues in seconds, but it seems far fetched on a forum. A lot of it was just experimenting and learning the API, but it is still no substitute for actual humans. Viewership is up, but participation remains unchanged.
 
You must log in or register to reply here.
Back
Top Bottom