AI-generated replies

Pollycat

PHK said:

booneruk said:

I just don't know what the motivation is, the user mentioned above (FlaatusGoat) seems to be the first to reply to a lot of newly created threads. Presumably they are pasting opening posts into an AI engine and relaying the generated text back in.

Why do this? It completely removes the human element. Is it for post building? is it for some strange kick? a misplaced desire to help?

Why would we need humans doing this? If it was a desirable feature (it really isn't!) the MSE devs could implement automatic AI responses to all new posts - a hideous thought indeed. I think something should be done to automatically detect and limit AI generated responses.

My suspicion is that some of these users are building up their post count before starting to spam. 

It is concerning there is no apparent plan in place to deal with all this AI nonsense being posted. 

In my experience they don't last long before being deleted/reported here:
Spam - Page 235 — MoneySavingExpert Forum

QrizB

booneruk said:

I've had a quick game of "Spot the human!" in their history. I think I ended up finding one post - needle in a haystack indeed.

That post is from a thread of theirs, regarding a fault with their vehicle. Every post of theirs in that thread looks/reads like a real human.

It's the replies from that account to other folks threads that have the hallmark of AI. It's almost as though the account's being used by two different people, one of whom thinks AI answers are useful. A child, perhaps, or someone else with loosely-supervised access to their PC/phone/tablet?

Grumpy_chap

Brie said:

I agree in a perfect world that should be done but I don't know how it can be managed without also deleting some regular content as well.  Many of us will quote a bit of something - basically showing our research.  

I am not sufficiently tech-savvy to know how to use AI, but I can sometimes see when AI has been used.

As for quoting content found within the interweb, that is common.  I have done it myself, like the examples in the following thread:
https://forums.moneysavingexpert.com/discussion/6576105/john-lewis-refusing-to-hand-over-the-item-i-bought/p4
I have stated that I located the content on the web and provided a link to the source.
I provided my narrative / interpretation of that source content, with my application to the case in discussion within the thread.
I invited comment on my interpretation

Simply pasting AI content is not helpful.  We all know how AI can be incorrect.  I think there was a recent case on the BBC News pages where news articles "gun-man has been arrested" were summarised by AI as "arrested gun-man killed himself" or something similar which added a significant inaccuracy.  
https://www.bbc.co.uk/news/articles/cd0elzk24dno

Perhaps, slightly odder, there were some posts a little while back where an individual (I cannot recall who) was responding to posts with something along the lines of "I put your question into AI and it gave the following answer......"  Also not really helpful, but at least signposted.

MSE_James

If you see a post you suspect to be AI-generated, please do report it using the 'report' button - the 'technical disruption' option is probably the best fit. We can and do regularly remove such posts and take appropriate action.  

Pollycat

MSE_James said:

If you see a post you suspect to be AI-generated, please do report it using the 'report' button - the 'technical disruption' option is probably the best fit. We can and do regularly remove such posts and take appropriate action.  

Thanks very much for your prompt response, James.

GingerTim

Thanks, James, really appreciate it. Is MSE doing anything to proactively prevent this stuff appearing in the first place?

MSE_JC

Hi @GingerTim,

We are exploring ways to help minimise the impact of disruptive AI content. However, it's important we tread carefully since (as touched on previously in this thread) we don't want to accidentally remove legitimate content or punish genuine users.

GingerTim

Thanks for the response, James, much appreciated!

marcia_

 It is difficult because I think there are probably many posters who genuinely think they are doing the right thing by C&P from AI 

[Deleted User]

A massive problem with people doing that is It depends on how you phrase a question, op responded to something is asked, I had been googling my question and got 2 opposite responses depending on how I phrased the question.