Should a member be allowed to make a post which is AI generated or AI mixed without disclosing such use of AI as part of the post?

[wil]

There are at least four issues here related to AI posts:

1) Required disclosure if a post is generated by AI. (I have posted ChatGPT answers to questions. In those posts I indicated conspicuously that I asked Chat GPT the question. I posted the verbatim question I asked, and then I posted the verbatim answer I received from ChatGPT.)

2) Are AI detectors reliable? (One post last week which was accused of being AI was rated 100% chance of AI by two detectors, and 55% chance by a third AI detector. I sure wouldn't want AI detectors to be adjudicating allegations against criminal defendants.)

3) Disciplinary action if a non-AI disclosed post is found by one or more AI detectors administered by the moderators or some other WBF deliberative body to be AI generated.

4) How should "mixed" AI be treated?

Just disclose when it’s AI and when not

 

[nirodha]

...how did you disconnect them from your brain?

That is the trick, for me to know and you to find our but don’t think about too much about it

 

[microstrip]

The problems we are addressing: https://originality.ai/blog/make-ai-writing-undetectable

These people are clear in their marketing

Used by Students & Writers Around the World​

 

[Mike Lavigne]

What rules?

whatever rules the forum adapts. which as of now is zero. just a groundswell opinions rejecting any use of AI, and maybe a bit of nuanced perspective too.. but that could change any time.

so i'm saying that however the forum decides to construct AI rules, translations combined with AI could help. if it's allowed.

 

[Ron Resnick]

Just disclose when it’s AI and when not

Disclosure is the easy question.

What about censorship of a possibly AI generated post which was not disclosed by the poster as such?

What about disciplining of a member who fails to disclose a post which an AI detector says was an AI generated post?

 

[Iamrael]

Is it possible for someone like Ted (Synergistic Research) or any others that appear to be relying on AI to generate content for their posts to simply ask CPChat for results that are not detectable by these “AI detectors”?

 

[DaveC]

Is it possible for someone like Ted (Synergistic Research) or any others that appear to be relying on AI to generate content for their posts to simply ask CPChat for results that are not detectable by these “AI detectors”?

I think that's already happened, or he's figured out how to alter it enough.

Once you're familiar with AI it's not hard to recognize, no detector needed. At least for now...

 

[Iamrael]

..since it doesn’t appear if anyone has asked…I will.
Ted, have you used AI to generate any or all of your posts?

 

[Rensselaer]

I am pretty much house bound so must get information on the outside world through tv and internet (newspapers in the UK IMHO are tabloids that regularly get sued for illegal phone tapping and just plain lies). In regards to television News, the usual news outlets are politically aligned so give slightly different perspectives of the exact same stories, usually at the same time (how is such organised?). The only exceptions being Euronews (boring) and Al Jazeera which still seems truthful and independent.

In regards to entertainment : I am so tired of algorithms limiting my Netflix and YouTube viewing to stories and genres that I have already seen. And sky tv, showing re-runs of programs that came out 50-70 years ago to get around royalties. Same YouTube. Instead of new and interesting material that would expose me to novel unexplored perspectives, I get flooded only with subjects I have already explored, and this is where AI comes in. On YouTube I find AI is being used to create fictional stories around subjects to make money.

I have a YouTube channel on which I play my records (recorded on my iPhone). If I advertised products on my channel (and had enough subscribers) I could make money from it, but I am not doing it for money but for the purpose of sharing so people who buy records can audition before buying.

I subscribe to videos of a civil rights lawyer who plays videos by victims of police crime shared with him and discusses the legal aspects. I learn about constitutional law and how different States tweak their laws to try and get around such. Because I subscribe to his YouTube channel the algorithms have flooded me with other seemingly related channels.

For a couple months now, I have noticed some AI-generated fake stories have been included in those alternative subject-related channels. They are immediately recognisable by two characteristics that give them away; first, the story starts in a pleasant story-telling voice; “John is a well-liked local factory worker with a lovely wife, Janice, and two happy sturdy school-age boys, Shaun and Thomas. One afternoon, while driving home from Bible study …” second, the photos, allegedly of John and his family, show different people each time it mentions John, Janice or the boys, smiling photos no-doubt borrowed from the internet. As the story proceeds, it goes on and on without getting to the event. You hear what a lovely town he lives in, what pretty scenery he was enjoying before his run-in with police, what he had for breakfast, ad nauseum.

Obviously a person has used AI to generate subject content for various subjects that his YouTube help has told him get a lot of hits. Once in place his AI slave just churns out the click-baits and fictitious stories, and he collects the money. Getting rich off lies (sound familiar?).

The only reason I joined the WBF was to learn from those with different experience to me. I did not join to read manufacturers peddling their snake oil with unproven claims or to read totally fictional AI content.

 

[microstrip]

(...) so i'm saying that however the forum decides to construct AI rules, translations combined with AI could help. if it's allowed.

The problem with translations is that AI translations (or text optimization with AI) is carried with an "intelligent" perspective of pleasing people and making it easy to read. Many times it will change the sense of subjective findings and expand such style in the forum.

See how AI "improved" my text:

The challenge with AI translations (or AI-assisted text optimization) is that they prioritize readability and user appeal. In doing so, they often alter the meaning of subjective findings and extend this stylistic approach throughout the forum.

A problem becomes a challenge ...

See what happens to your very simple sentence ;

So, what I'm saying is that regardless of how the forum chooses to establish AI rules, AI-assisted translations could be beneficial—if permitted.

Was this exactly what you wanted to say?

 

[Ron Resnick]

The problem with translations is that AI translations (or text optimization with AI) is carried with an "intelligent" perspective of pleasing people and making it easy to read. Many times it will change the sense of subjective findings and expand such style in the forum.

See how AI "improved" my text:

The challenge with AI translations (or AI-assisted text optimization) is that they prioritize readability and user appeal. In doing so, they often alter the meaning of subjective findings and extend this stylistic approach throughout the forum.

A problem becomes a challenge ...

See what happens to your very simple sentence ;

So, what I'm saying is that regardless of how the forum chooses to establish AI rules, AI-assisted translations could be beneficial—if permitted.

Was this exactly what you wanted to say?

This raises an interesting question, Francisco. What if a member free drafts (writes something completely by himself/herself) but then runs it through AI to clean it up and to fix grammar and typos? This sounds like a high-tech version of a spell-checker.

My answer to this question: I do not see how this "reverse" (checker) use of AI should, or even can, be proscribed. This then takes us to so-called "mixed AI" which appears impossible to untangle.

This is why efforts to ban AI in the WBF context are misguided or extremely prone to arbitrary or biased censorship. As a libertarian I support (only) required (i.e., voluntary) disclosure. Going beyond disclosure, and attempting to adjudicate posts by using AI detectors and then imposing discipline based thereon, is a bad and unworkable idea.

 

[Al M.]

Once you're familiar with AI it's not hard to recognize, no detector needed.

So what are the telltale signs, in your experience?

 

[microstrip]

This raises an interesting question, Francisco. What if a member free drafts (writes something completely by himself/herself) but then runs it through AI to clean it up and to fix grammar and typos?

My answer to this question: I do not see how this use of AI should, or even can, be proscribed. This gets us to so-called "mixed AI" which appears impossible to untangle.

This is why efforts to ban AI are misguided or extremely prone to arbitrary censorship.

Surely - it was the sense of my previous posts. We have to coexist with it, but this should motivate members to understand and debate it.

But IMO content "generated by AI" should be banned, as well as members posting it. YMMV.

 

[Mike Lavigne]

The problem with translations is that AI translations (or text optimization with AI) is carried with an "intelligent" perspective of pleasing people and making it easy to read. Many times it will change the sense of subjective findings and expand such style in the forum.

See how AI "improved" my text:

The challenge with AI translations (or AI-assisted text optimization) is that they prioritize readability and user appeal. In doing so, they often alter the meaning of subjective findings and extend this stylistic approach throughout the forum.

A problem becomes a challenge ...

See what happens to your very simple sentence ;

So, what I'm saying is that regardless of how the forum chooses to establish AI rules, AI-assisted translations could be beneficial—if permitted.

AI can be harmless, or insidious. it's a fine line even when innocently applied. where does it step off into something else?

Was this exactly what you wanted to say?

it is close enough to be considered the same. but where does it go 'off'? i don't know. better to not go there.

honestly i have not really thought about it much. most media i don't believe anyway. face value has little value. and am always looking for alternate views for things i care about.

 

[Steve williams]

This raises an interesting question, Francisco. What if a member free drafts (writes something completely by himself/herself) but then runs it through AI to clean it up and to fix grammar and typos? This sounds like a high-tech version of a spell-checker.

My answer to this question: I do not see how this "reverse" (checker) use of AI should, or even can, be proscribed. This then takes us to so-called "mixed AI" which appears impossible to untangle.

This is why efforts to ban AI in the WBF context are misguided or extremely prone to arbitrary or biased censorship. As a libertarian I support (only) required (i.e., voluntary) disclosure. Going beyond disclosure, and attempting to adjudicate posts by using AI detectors and then imposing discipline based thereon, is a bad and unworkable idea.

The poll results speak otherwise. Let’s keep your politics out of this

 

[Ron Resnick]

The poll results speak otherwise.

How do the poll results speak otherwise?

 

[Mike Lavigne]

regarding Artificial Intelligence and hifi forums and myself; i enjoy the process of thinking and crafting a post. in an environment where i'm comfortable. anything which corrupts the communication or validity of real world back and forth thoughts is bad. my spider sense recognizes AI content as not coming from a good place, whether i actually assign the 'AI' label to it or not.

kinda like MFSL records i bought over a 10 year period where i never played them more than the first couple cuts of one side. they just did not capture me. preferred my earlier pressings. then later learned they were done with digital masters. it's how our minds work.

and posters who have used AI have not fared well overall in my smell test. and once exposed as false, hard to un-know it.

 

[DaveC]

So what are the telltale signs, in your experience?

Both writing and photos have telltale signs. I don't want to reinvent the wheel here so I won't go into detail, if you want details you can ask AI, lol... but I think right now, both AI-generated photos and writing are pretty obvious. It takes looking at them and analyzing them to get a feel for it. Like listening critically to audio, recognizing AI cues is a learned skill. But AI learns and it will get more and more difficult to tell.

 

[DaveC]

This raises an interesting question, Francisco. What if a member free drafts (writes something completely by himself/herself) but then runs it through AI to clean it up and to fix grammar and typos? This sounds like a high-tech version of a spell-checker.

My answer to this question: I do not see how this "reverse" (checker) use of AI should, or even can, be proscribed. This then takes us to so-called "mixed AI" which appears impossible to untangle.

This is why efforts to ban AI in the WBF context are misguided or extremely prone to arbitrary or biased censorship. As a libertarian I support (only) required (i.e., voluntary) disclosure. Going beyond disclosure, and attempting to adjudicate posts by using AI detectors and then imposing discipline based thereon, is a bad and unworkable idea.

Ron, if this is the case all forums are now useless and a vast majority of real humans will stop participating. As a forum owner, I have a hard time understanding your views here, you are advocating for the obsolescence of your own forum.

I think it is critical to be able to know whether content is AI or human generated just in general. If not, it's not just forums that are obsolete, our entire education system is now compromised to the point a degree means nothing.

I'd love to hear a reasonable argument where not being able to differentiate AI and humans results in any other outcome that what I stated above.

 

[Ron Resnick]

Ron, if this is the case all forums are now useless and a vast majority of real humans will stop participating.

Huh? If I see somebody writing here with what appears to be AI I simply will put the person on "ignore." Why would I want to read AI crap on this hobby forum?

you are advocating for the obsolescence of your own forum.

Huh?

I think it is critical to be able to know whether content is AI or human generated just in general.

This seems like another way overbroad comment. I don't think this is a matter of generality.

In some situations such as seeking accurate and intelligible information I may not care if it's AI or not. On my own audio forum for the hobby I love I have no interest in reading AI.

If not, it's not just forums that are obsolete

Huh?

our entire education system is now compromised to the point a degree means nothing.

I'd love to hear a reasonable argument where not being able to differentiate AI and humans results in any other outcome that what I stated above.

Hi Dave,

I truly do not understand most of your post, and your questions. I certainly don't agree with your universal declarations. I do not know how you arrived at those apocalyptic conclusions from my post to which you are replying.

Please ask me a clear, specific question.

Thank you.