I came across an example today of a member using chatGPT to generate a post. They were open about having done this, first time, but then added a second post, also AI generated, but which didn't make this clear. It saved them time, apparently.
This bothers me. I'm not sure I want to be reading posts written by an AI, especially if the authorship is not clear.
I looked to see if there's a forum rule about this, but I can't find anything.
This bothers me. I'm not sure I want to be reading posts written by an AI, especially if the authorship is not clear.
I looked to see if there's a forum rule about this, but I can't find anything.
URL to the post please.
Not offical policy but…
A subject in careful consideration and concern in the back-room.
Acknowledging that a post is AI generated might make it acceptable, they are usually so they can be laughed at.
An AI (idiot Savant a better name but human savants are much, much more accurate than ChatGPT) generated post with no acknowledgement is a goner.
dave
Not offical policy but…
A subject in careful consideration and concern in the back-room.
Acknowledging that a post is AI generated might make it acceptable, they are usually so they can be laughed at.
An AI (idiot Savant a better name but human savants are much, much more accurate than ChatGPT) generated post with no acknowledgement is a goner.
dave
Last edited:
I agree with you Ianbo.
This topic is being actively discussed by the site staff. We have been working recently to improve the “front gate” of the site as we saw AI powered spambots starting to leak through.
I and I’m sure most members don’t want to see AI generated content here that is being passed off as human content.
At a minimum it should be clearly marked as such, and we might create a specific BBcode tag to use to do so. Then it’s clear what is what.
At a maximum all AI content would be banned.
There is some nuance when it comes to using it for translation or spelling/grammar/proofreading. Every text editor will have essentially an AI powered auto-correct very shortly.
This topic is being actively discussed by the site staff. We have been working recently to improve the “front gate” of the site as we saw AI powered spambots starting to leak through.
I and I’m sure most members don’t want to see AI generated content here that is being passed off as human content.
At a minimum it should be clearly marked as such, and we might create a specific BBcode tag to use to do so. Then it’s clear what is what.
At a maximum all AI content would be banned.
There is some nuance when it comes to using it for translation or spelling/grammar/proofreading. Every text editor will have essentially an AI powered auto-correct very shortly.
This topic caught my eye, since it's relevant to my experience over the last few weeks with this community.
Like a lot of folks here, I'm an amateur. I'm learning the basics, doing my homework, etc.
I've found Chat GPT (specifically the 'agent' I customized to help me with speaker design and electrical engineering) to be an incredible boon to how I engage with this forum.
It's helped me jump into conversations on topics I would've normally watched from the sidelines.
I've considered using it to post questions, too – just so that my amateur fumblings don't get ignored or misunderstood.
But I haven't, because unfortunately everything invariably comes out like the robotic drivel in the post @ianbo shared. I do appreciate that they were transparent about it and playing around with it.
In theory, we can (and likely will) unkowningly have conversations with AI generated copy-pasta on this forum. How is that different from any other deceptive behavior?
Like a lot of folks here, I'm an amateur. I'm learning the basics, doing my homework, etc.
I've found Chat GPT (specifically the 'agent' I customized to help me with speaker design and electrical engineering) to be an incredible boon to how I engage with this forum.
It's helped me jump into conversations on topics I would've normally watched from the sidelines.
I've considered using it to post questions, too – just so that my amateur fumblings don't get ignored or misunderstood.
But I haven't, because unfortunately everything invariably comes out like the robotic drivel in the post @ianbo shared. I do appreciate that they were transparent about it and playing around with it.
In theory, we can (and likely will) unkowningly have conversations with AI generated copy-pasta on this forum. How is that different from any other deceptive behavior?
Keep in mind that it seems that is right a bit over 50% of the time, so you will get as much misinformation as information.
dave
The linked post looks much more like someone used a line or two request to generate a large amount of 'content'. Saving them from having to do the research and put it into their own words.
I'm not generally a fan of really long posts but often if someone put the time and effort into writing it there may be something good there. Why visit a forum if I can just have AI generate 'posts' about everything I want to read about? Heck, I just generated several pages arguing for and against AI posts on forums.
I view this forum like chatting with friends and acquaintances about electronics and audio. I don't want to have a conversation with someone that is just going to hold up pages from a book or magazine for their part of the interaction.
Do we openly accept other forms of deceptive behavior?
Using it as an education source like you did is one thing, using it to bypass actually processing the information found online yourself is another.
For myself I would much more likely read a long post someone composed on their own and would skip anything that starts with "According to ChatGPT:"
IMO it is worse to read lots of titles or threads regarding speaker cables or capacitors as they are supposed to affect sound, the vast majority with little or no knowledge nor laboratory tests.
To wit, I could have proofed this with GPT to clarify. I didn't intend to inflate the (IMO) inevitability of AI tainted posts with the (IMO) deceptive nature of not disclosing the use of AI.
I'm sure GPT would help me find the perfect aphorism to add here... too bad I can't put my finger on it
Very soon our lives will be flooded with instant and easy access to soulless (but very human sounding) AI generated answers. The web will be flooded with AI generated spam trained itself on AI generated spam. You won’t be able to trust anything you read, and so trust will become everything.
In those days, what will diyAudio be? A community of humans sharing their human experience. A source of information, community and camaraderie from and with humans.
Don’t see much place in there for AI content. Human content and interaction and community will be what makes diyAudio special and unique. An island amongst the chaos.
I can see there being a future need or benefit to being about to quote AI content. But it should be clearly marked and quarantined as such.
In those days, what will diyAudio be? A community of humans sharing their human experience. A source of information, community and camaraderie from and with humans.
Don’t see much place in there for AI content. Human content and interaction and community will be what makes diyAudio special and unique. An island amongst the chaos.
I can see there being a future need or benefit to being about to quote AI content. But it should be clearly marked and quarantined as such.
When everyone just relies on AI for their information, who will generate new information for AI to feed on?
The plus there is those are easy to spot and skip if you want to.
There are AI content detectors for those having doubts about a text. One of those is
https://www.scribbr.com/ai-detector/
From there to an app, "proofreading" every post in a forum can't take long if just to prevent flooding forums with AI (Artificial Idiot) posts.
On that site, the above shows a 7% AI generated probability.
https://www.scribbr.com/ai-detector/
From there to an app, "proofreading" every post in a forum can't take long if just to prevent flooding forums with AI (Artificial Idiot) posts.
On that site, the above shows a 7% AI generated probability.
I did a test a while ago with about ten different detectors on one post I found somewhere on this forum. They all came up with different results, from 0% to 100% AI generated. Totally unreliable.
The forum thrives on member-generated input.
Hugo
I took UseNet posts to test the mentioned detector in different languages and subjects. The ones I suspected to have major AI generated content were shown with over 60% probability of that whereas human-originated texts were no higher than 20%. AFAIK that's good enough so I did not check the competition.