How do you handle members who use AI in their posts?

[island]

Seeing an increase in the number of posts that have AI parts in them. Still not huge, but I suspect it will become an increasing thing to think about.

Should members who use AI to compose their post be required to put the AI generated parts in some kind of AI-quote tag?

In the longer term, as people start using AI more and more to generate all their writing, how will this affect forums and the human connection?

 

[Sim]

Clear policies is the first step - let your members know what is expected and what is not permitted.

While I do have a policy forbidding any AI generated content, sometimes it is useful to have an AI summary of something - provided that it is clearly documented as being AI generated. I was actually thinking today that some type of "AI-quote" tag might be useful for this situation.

 

[Ricsca]

It's already a miracle that users post in forums. I'd say don't overwhelm them with unnecessary rules...
In the end, it's still a discussion or a post, and it's definitely more accurate since it's written by artificial intelligence based on what the user wanted to say...

 

[Mr Lucky]

I was actually thinking today that some type of "AI-quote" tag might be useful for this situation.

Although it’s easy enough with custom bbcode I think this would make a god suggestion.

and it's definitely more accurate since it's written by artificial intelligence based on what the user wanted to say...

Unless it isn’t. It are you saying ai never hallucinates or regurgitates false information?

 

[Forsaken]

Only ~community I have will likely have an issue with people posting blatant AI slop, and only one has so far and they basically got mass ignored because of it.

On other forums, I am just ignoring people who just blatantly post using AI, because they almost never provide any value to discussions in the first place. Until XF has guidelines for AI usage on add-ons, I am also ignoring most devs I think are just vibe coding add-ons as well.

 

[StarArmy]

I don't mind AI-assisted posts too much, as long as the post is of high quality. But I don't like the blatant use of AI where the user obviously didn't review/edit what they posted before they posted it. Like @Forsaken said, the key word is blatant. If you're thinking "this is AI talking" when reading instead of "this is the user responding" then that's a good sign that the post is "low effort" or "low value." A basic rule to use for this might be:

• Be yourself:
  • We want to hear from you! That's why we're on this forum not asking Gemini, etc; therefore, make sure it's still you talking in the posts. Using some AI to help write is permitted, but blatantly low-effort robot posts may be removed by staff.

 

[Ricsca]

Unless it isn’t. It are you saying ai never hallucinates or regurgitates false information?

You can make mistakes that are certainly 99% less than the mistakes a user can make...

 

[ES Dev Team]

Our rule:



..and yes, we enforce it

 

[sub_ubi]

LLMs are an internet revolution, they've already created the best movies and books and games, shouldn't you be celebrating each new AI post as a gift from the singularity?

 

[Forsaken]

You can make mistakes that are certainly 99% less than the mistakes a user can make...

A person making a mistake can take accountability for their mistake and hopefully learn from them; AI cannot take accountability (and the companies behind AI models avoid accountability like the plague) and does not learn from it's mistakes. Comparing the two is nonsensical.

 

[Mr Lucky]

You can make mistakes that are certainly 99% less than the mistakes a user can make...

Less or fewer?

 

[Taylor J]

You can make mistakes that are certainly 99% less than the mistakes a user can make...

It would help to remember that LLM's are just statistics under the hood and are all setup to not disappoint the end user I.E: If an llm can't find you a correct answer to your question they'll give you any answer as to never come back with their hands empty, unless specified that you are okay with that.