Donate to Remove ads

Got a credit card? use our Credit Card & Finance Calculators

Thanks to waiting4godot,yyuryyub,88V8,Anonymous,Fluke, for Donating to support the site

Artificial Members

Formerly "Lemon Fool - Improve the Recipe" repurposed as Room 102 (see above).
mc2fool
Lemon Half
Posts: 7693
Joined: November 4th, 2016, 11:24 am
Has thanked: 7 times
Been thanked: 2957 times

Artificial Members

#631412

Postby mc2fool » December 3rd, 2023, 2:30 pm

No, I'm not talking about the Ann Summers Rabbits ... :D

There's been a few posts recently which really smack of having been written by ChaptGPT or some other generative AI. Whether the members are an AI bot that has, somehow, managed to sign itself up and is posting autonomously, or whether the members are humans that are using AI to write posts isn't clear.

Nor is it clear, to me at least, what should, or indeed, could be done about it, beyond the after-the-fact usual reporting and moderation mechanisms. I'm not sure if it's really a "Site Issue" (yet), so I'm posting this under the "General Chat" remit of this board, rather than as an issue or "Complaint".

Here's the two from today that triggered me to write here. Discuss. ;)

This one in the Taxes board, reviving a thread on Self Assessment - Self-employment section that was previously last active in 2018:

"Completing the self-employment section was an insightful process. It allowed me to reflect on my entrepreneurial journey, assess financial accomplishments, and identify areas for growth. This self-assessment is a valuable tool for setting future goals and enhancing business strategies."
viewtopic.php?p=631350#p631350

And this one in DAK, reviving a thread on Why not fill hot water bottle from hot tap? that was previously last active in May 2023:

"Filling a hot water bottle from the hot tap is generally not recommended due to the potential risks associated with hot tap water. Hot tap water can sometimes come from a system that may not be well-suited for consumption or use in a hot water bottle. Here are a few reasons why using hot tap water may not be advisable:
Temperature Fluctuations: The idea of using a heating pad staying warm the paste and other components in cold weather is ingenious. Hot tap water temperatures can vary, and in some cases, it may be too hot. Extremely hot water can cause burns or scalds, especially when in direct contact with the skin.
Mineral Content: Hot tap water may contain a higher concentration of minerals, including metals like lead. Using such water in a hot water bottle can lead to the release of these minerals into the water, and if the water comes into contact with the skin, it may pose health risks.
Bacterial Growth: Hot tap water might be stored in a water heater tank, and over time, bacteria can grow in warm environments. Using this water in a hot water bottle without proper precautions could introduce bacteria to the bottle.
"
viewtopic.php?p=631404#p631404

chas49
Lemon Quarter
Posts: 1875
Joined: November 4th, 2016, 10:25 am
Has thanked: 207 times
Been thanked: 442 times

Re: Artificial Members

#631415

Postby chas49 » December 3rd, 2023, 2:42 pm

I just deleted my post replying to the hot water bottle thread! I note that 'paultate' joined today and posted that rather incoherent message. I have deleted the post, and reported to admins as a likely spammer.

mc2fool
Lemon Half
Posts: 7693
Joined: November 4th, 2016, 11:24 am
Has thanked: 7 times
Been thanked: 2957 times

Re: Artificial Members

#631423

Postby mc2fool » December 3rd, 2023, 3:13 pm

chas49 wrote:I just deleted my post replying to the hot water bottle thread! I note that 'paultate' joined today and posted that rather incoherent message. I have deleted the post, and reported to admins as a likely spammer.

Actually there's only one sentence that's syntactically incoherent; the rest of it was fine in that respect and at least coherent enough for you to reply to and point out the logic errors in (which I saw). ;)

ReformedCharacter
Lemon Quarter
Posts: 3102
Joined: November 4th, 2016, 11:12 am
Has thanked: 3515 times
Been thanked: 1501 times

Re: Artificial Members

#631427

Postby ReformedCharacter » December 3rd, 2023, 3:29 pm

mc2fool wrote:
chas49 wrote:I just deleted my post replying to the hot water bottle thread! I note that 'paultate' joined today and posted that rather incoherent message. I have deleted the post, and reported to admins as a likely spammer.

Actually there's only one sentence that's syntactically incoherent; the rest of it was fine in that respect and at least coherent enough for you to reply to and point out the logic errors in (which I saw). ;)

ChatGPT, I guess.

RC

88V8
Lemon Half
Posts: 5628
Joined: November 4th, 2016, 11:22 am
Has thanked: 3911 times
Been thanked: 2478 times

Re: Artificial Members

#631446

Postby 88V8 » December 3rd, 2023, 5:08 pm

ReformedCharacter wrote:
mc2fool wrote:Actually there's only one sentence that's syntactically incoherent; the rest of it was fine in that respect...

ChatGPT, I guess.

At least it can spell. Some sellers on eBay would find it beneficial ;)

V8

mc2fool
Lemon Half
Posts: 7693
Joined: November 4th, 2016, 11:24 am
Has thanked: 7 times
Been thanked: 2957 times

Re: Artificial Members

#631529

Postby mc2fool » December 3rd, 2023, 11:06 pm


Lootman
The full Lemon
Posts: 18166
Joined: November 4th, 2016, 3:58 pm
Has thanked: 609 times
Been thanked: 6311 times

Re: Artificial Members

#631530

Postby Lootman » December 3rd, 2023, 11:10 pm

mc2fool wrote:And maybe another? viewtopic.php?p=631495#p631495

Xfool may be many things. But a bot, not.

mc2fool
Lemon Half
Posts: 7693
Joined: November 4th, 2016, 11:24 am
Has thanked: 7 times
Been thanked: 2957 times

Re: Artificial Members

#631531

Postby mc2fool » December 3rd, 2023, 11:13 pm

Lootman wrote:
mc2fool wrote:And maybe another? viewtopic.php?p=631495#p631495

Xfool may be many things. But a bot, not.

Ha ha. The link, of course, is to post 631495 by arhamarshad068.

bungeejumper
Lemon Half
Posts: 7921
Joined: November 8th, 2016, 2:30 pm
Has thanked: 2768 times
Been thanked: 3855 times

Re: Artificial Members

#631610

Postby bungeejumper » December 4th, 2023, 12:55 pm

I've reported a few more over the last few months. Dispensing truisms with such elaborate care that my inner ear instinctively hears them in a tinny Robbie the Robot voice.

I rather enjoyed this recent exchange on an AI discussion forum: https://community.openai.com/t/i-just-h ... -it/316377
User: Just wanted to make sure. “Better safe than sorry” is an appropriate truism… (truism, is that the word im looking for?)

ChatGPT: Yes, “truism” is the correct term. A truism is a statement that is obviously true and says nothing new or interesting. The phrase “better safe than sorry” is indeed a truism as it’s a widely accepted fact that it’s better to be cautious than to take unnecessary risks. It’s often used to justify precautionary actions or safety measures. Your usage of the term is correct.

:lol: But the worrying thing is that they'll probably get less obvious than this.....

BJ

XFool
The full Lemon
Posts: 12329
Joined: November 8th, 2016, 7:21 pm
Been thanked: 2521 times

Re: Artificial Members

#631616

Postby XFool » December 4th, 2023, 1:18 pm

So what would be the point? Testing, to see if they are picked up as AI or not?

csearle
Lemon Quarter
Posts: 4695
Joined: November 4th, 2016, 2:24 pm
Has thanked: 4705 times
Been thanked: 2027 times

Re: Artificial Members

#631718

Postby csearle » December 4th, 2023, 9:38 pm

Lootman wrote:
mc2fool wrote:And maybe another? viewtopic.php?p=631495#p631495

Xfool may be many things. But a bot, not.
The link points to a different poster (memberlist.php?mode=viewprofile&u=12255). C.

BullDog
Lemon Quarter
Posts: 2342
Joined: November 18th, 2021, 11:57 am
Has thanked: 1876 times
Been thanked: 1158 times

Re: Artificial Members

#631764

Postby BullDog » December 5th, 2023, 8:31 am

Yes, I noticed posts looking suspiciously like bots a few months ago and reported them to moderators. It's most certainly happening here.

XFool
The full Lemon
Posts: 12329
Joined: November 8th, 2016, 7:21 pm
Been thanked: 2521 times

Re: Artificial Members

#632772

Postby XFool » December 9th, 2023, 11:09 am

XFool wrote:So what would be the point? Testing, to see if they are picked up as AI or not?

To suggest an answer to my own question: What might be called 'Turing Testing' ?

That is, it's one way of using the Internet itself to conduct testing to see if your AI Bot can, in practice, effectively pass the Turing Test. A success being not to be 'outed' as a bot.

Lanark
Lemon Quarter
Posts: 1286
Joined: March 27th, 2017, 11:41 am
Has thanked: 582 times
Been thanked: 558 times

Re: Artificial Members

#632782

Postby Lanark » December 9th, 2023, 11:50 am

If spammers can find a way to have an AI robot post multiple times without being noticed, then it starts to get much easier to sneak in advertising links.

XFool
The full Lemon
Posts: 12329
Joined: November 8th, 2016, 7:21 pm
Been thanked: 2521 times

Re: Artificial Members

#633626

Postby XFool » December 13th, 2023, 9:44 am

I'm starting to wonder is somebody is working on a PhD project hereabouts. Perhaps an AI or psychology PhD?

Thesis: "What happens if real humans, in the wild, are faced with the prospect - "I've had a great idea. Why don't we perform the Turing Test, right here!"

:?

csearle
Lemon Quarter
Posts: 4695
Joined: November 4th, 2016, 2:24 pm
Has thanked: 4705 times
Been thanked: 2027 times

Re: Artificial Members

#633829

Postby csearle » December 13th, 2023, 9:48 pm

Back in the eighties I was writing a PhD on AI at Plymouth under a certain Prof. Baker. Only for six months though. It wasn't really what I'd imagined and then I met a German girl. C.

CliffEdge
Lemon Quarter
Posts: 1553
Joined: July 25th, 2018, 9:56 am
Has thanked: 449 times
Been thanked: 433 times

Re: Artificial Members

#633864

Postby CliffEdge » December 14th, 2023, 1:21 am

csearle wrote:Back in the eighties I was writing a PhD on AI at Plymouth under a certain Prof. Baker. Only for six months though. It wasn't really what I'd imagined and then I met a German girl. C.

That's very reassuring.

jfgw
Lemon Quarter
Posts: 2532
Joined: November 4th, 2016, 3:36 pm
Has thanked: 1088 times
Been thanked: 1145 times

Re: Artificial Members

#633930

Postby jfgw » December 14th, 2023, 12:12 pm

mc2fool wrote:There's been a few posts recently which really smack of having been written by ChaptGPT or some other generative AI. Whether the members are an AI bot that has, somehow, managed to sign itself up and is posting autonomously...


ChatGPT denies it.

Me: "Do you post messages on The Lemon Fool?"

ChatGPT: "As an AI, I do not have the capability to post messages or interact with forums such as The Lemon Fool. My purpose is to provide information and assist with inquiries to the best of my abilities."

That does not rule out a human (or non-human) accomplice. It does not rule out ChatGPT telling lies either, or another AI entity signing up.

Me: "Does anyone use you to answer questions on The Lemon Fool?"

ChatGPT: "failed to get data, error status 503"

Is it hiding something? I asked the same question again,

ChatGPT: "failed to get data, error status 500"

Third time lucky? I asked again,

ChatGPT: "No, I am an AI language model developed by OpenAI and I am not capable of browsing the internet or accessing specific websites like The Lemon Fool. My purpose is to assist users in generating human-like text based on the prompts provided to me."

It is skirting the issue. I don't trust it. Arthur C. Clarke and Stanley Kubrick warned us about this.


Julian F. G. W.

chas49
Lemon Quarter
Posts: 1875
Joined: November 4th, 2016, 10:25 am
Has thanked: 207 times
Been thanked: 442 times

Re: Artificial Members

#633934

Postby chas49 » December 14th, 2023, 12:18 pm

I asked ChatGPT this:

me wrote:Can you wite a post for a forum about how an AI can post on a forum?


It replied:

ChatGPT wrote:Subject: Embracing the Future: AI Forum Contributions

Hey [Forum Name] Community,

I hope this message finds you all well! Today, I want to share something exciting with you all — the emergence of Artificial Intelligence (AI) and its potential to actively participate in our forum discussions.

As technology continues to advance, AI has made significant strides in natural language processing, enabling it to engage in conversations much like humans. You might be wondering, "How can an AI contribute meaningfully to our forum?" Well, here are some key points to consider:

Information Retrieval and Synthesis:
AI models, like the one generating this post, can sift through vast amounts of information quickly. This ability allows them to provide concise and relevant answers by synthesizing data from various sources.

24/7 Availability:
Unlike humans, AI doesn't need sleep or take breaks. This means it can be available around the clock, offering timely responses and assistance to forum members from different time zones.

Language Understanding:
Advanced AI models understand context, sentiment, and nuances in language. This enables them to participate in discussions more intelligently, providing thoughtful responses that align with the forum's atmosphere.

Community Assistance:
AI can be programmed to assist users by answering frequently asked questions, guiding newcomers, and offering solutions to common issues. This can enhance the overall user experience and make the forum more accessible.

Diverse Perspectives:
AI models can be trained on a wide range of data, allowing them to bring diverse perspectives to discussions. This can enrich conversations and expose forum members to different viewpoints.

However, it's important to note that while AI can be a valuable addition to our community, it should complement, not replace, human interactions. The human touch, empathy, and creativity are integral aspects of our forum, and AI should enhance these qualities rather than diminish them.

As we explore the possibilities of AI engagement on our forum, your feedback is crucial. Share your thoughts on how you envision AI contributing positively to our community, and let's embark on this technological journey together.

Looking forward to hearing your thoughts and ideas!

Best regards,

[Your Forum Username]


There's certainly some similarity between this language and the posting style which has engendered this thread.

What do you most intelligent and esteemed friends on this forum think? :)

chas49
Lemon Quarter
Posts: 1875
Joined: November 4th, 2016, 10:25 am
Has thanked: 207 times
Been thanked: 442 times

Re: Artificial Members

#633935

Postby chas49 » December 14th, 2023, 12:21 pm

And then I asked:

Me: How would you decide if a forum post had been written by an AI?

ChatGPT: Detecting whether a forum post has been written by an AI can be challenging, especially as AI language models like mine (GPT-3.5) are designed to generate human-like text. However, there are a few potential indicators that might suggest a post was written by an AI:

Consistency:
AI models are consistent in their writing style, and they don't have personal experiences or emotions. If a post consistently lacks personal anecdotes, emotions, or inconsistencies typically found in human communication, it might raise suspicions.

Complexity and Depth:
AI models can generate highly complex and detailed content on a wide range of topics. If a post demonstrates an unusually high level of complexity, uses advanced vocabulary consistently, or covers a diverse range of subjects without any signs of struggle, it could be a hint.

Unusual or Unique Responses:
AI models might generate responses that are unconventional or too perfect. If a post consistently offers flawless, almost "ideal" answers without any human touch, it might suggest AI involvement.

Response Time:
AI models can generate responses very quickly. If a post is submitted almost instantly after a question or comment is posted, and it's consistently quick in responses, it could indicate automated generation.

Lack of Personal Experience:
AI lacks personal experiences and emotions. If a post lacks personal anecdotes, opinions, or emotions, it might suggest it was written by an AI.

Repetitiveness:
AI models may exhibit repetitive patterns in their responses. If a post seems to reuse phrases or structures in a way that is not typical of human communication, it could be a sign of automated content generation.

Unusual Requests or Prompts:
AI models might respond strangely to unusual or unexpected requests or prompts. If a post doesn't handle these situations appropriately or exhibits confusion where a human likely wouldn't, it might be indicative of AI.

It's important to note that these indicators are not foolproof, and some highly advanced AI models can mimic human writing to a remarkable extent. Additionally, human users may also exhibit some of the characteristics listed above. Therefore, it's often challenging to definitively determine if a post was written by an AI based solely on its content.


Return to “Room 102 - Site Issues, Complaints & General Chat”

Who is online

Users browsing this forum: No registered users and 11 guests