‘I realised I’d been ChatGPT-ed into bed’: how ‘Chatfishing’ made finding love on dating apps even weirder

[Innula Zenovka]

‘I realised I’d been ChatGPT-ed into bed’: how ‘Chatfishing’ made finding love on dating apps even weirder -- lengthy Guardian piece on the subject.

Where once people were duped by soft-focus photos and borrowed chat-up lines, now they have to watch out for computer-generated charm. But it’s one thing to use a witty phrase – another thing entirely to build a whole fake persona …

“From the beginning he was asking very open-ended questions, and that felt refreshing,” says Rachel. One early message from her match read: “I’ve been reading a bit about attachment styles lately, it’s helped me to understand myself better – and the type of partner I should be looking for. Have you ever looked at yours? Do you know your attachment style?” “It was like he was genuinely trying to get to know me on a deeper level. The questions felt a lot more thoughtful than the usual, ‘How’s your day going?’” she says.

While my interactions with ChatGPT are all about coding, recipes or stuff I can't be bothered to Google, I must say I find its habit of following everything up with open-ended questions intended to keep the conversation going can become very creepy and annoying.

 

[Soen Eber]

What keeps me off of AI is the knowledge that a lot of it is created by profit-motivated weirdos who want to maximize my engagement and my attention, and that just is not healthy. It's also why I stay off of most social media. And six fingered hands, somehow for some I just know Musk is behind the hands.

 

[GoblinCampFollower]

Specifically for the purpose of a hookup, it wouldn't bother me if some guys used AI's to sound more charming than they really are. But obviously that is a horrible basis for a long term relationship. It's even worse if it's to scam people out of money. In other words, charm and proper grammar are good, but deception is bad.

What keeps me off of AI is the knowledge that a lot of it is created by profit-motivated weirdos who want to maximize my engagement and my attention, and that just is not healthy. It's also why I stay off of most social media. And six fingered hands, somehow for some I just know Musk is behind the hands.

YEP, this is a lot of why I don't use AI's very heavily. And usually when I do use it it's at work and I tell everyone I used AI because I'm supposed to use it for the appearance of "innovation!!!!" lol.

 

[Noodles]

What keeps me off of AI is the knowledge that a lot of it is created by profit-motivated weirdos who want to maximize my engagement and my attention, and that just is not healthy. It's also why I stay off of most social media. And six fingered hands, somehow for some I just know Musk is behind the hands.

I have used it for roughing out some story ideas. And you basically have to kill the character its controlling to makemit stop, and even then it will keep going on and on. Only like once did I get it to give me a "The End" moment.

 

[Argent Stonecutter]

What keeps me off of AI is the knowledge that a lot of it is created by profit-motivated weirdos who want to maximize my engagement and my attention

Also by academic weirdos who want to maximize it. Because a huge percentage of AI research for the past 60 years has been dedicated to "beating the Turing test", which inherently involved maximizing the subject's attention so you can gaslight them into thinking that a piece of software is a real boy.

And the Turing test was an offhand thought experiment designed to help people over the hurdle of accepting the possibility of machine reasoning. Ironic that it turns out that it doesn't actually need any reasoning to pass.

 

[GoblinCampFollower]

Also by academic weirdos who want to maximize it. Because a huge percentage of AI research for the past 60 years has been dedicated to "beating the Turing test", which inherently involved maximizing the subject's attention so you can gaslight them into thinking that a piece of software is a real boy.

And the Turing test was an offhand thought experiment designed to help people over the hurdle of accepting the possibility of machine reasoning. Ironic that it turns out that it doesn't actually need any reasoning to pass.

Yeah, I've thought about the Turing test a lot and how it was a good thought experiment but maybe much easier than expected and still a far cry from true intelligence. I guess we don't really expect people to say something brilliant during normal conversation.

 

[Innula Zenovka]

What keeps me off of AI is the knowledge that a lot of it is created by profit-motivated weirdos who want to maximize my engagement and my attention, and that just is not healthy.

How does that work, though? The longer ChatGPT keeps me talking by suggesting follow-up questions I might like it to answer, or additional tasks I might like it to undertake, the greater the marginal cost to OpenAI, so I would have thought they'd want to answer my question without inviting me to stick around and ask them something else. And they're not selling advertising space, or at least not at the moment, so they can't monetise my attention that way.

 

[Soen Eber]

How does that work, though? The longer ChatGPT keeps me talking by suggesting follow-up questions I might like it to answer, or additional tasks I might like it to undertake, the greater the marginal cost to OpenAI, so I would have thought they'd want to answer my question without inviting me to stick around and ask them something else. And they're not selling advertising space, or at least not at the moment, so they can't monetise my attention that way.

They will eventually. "The first one is always free" -every drug dealer.

 

[Bartholomew Gallacher]

While my interactions with ChatGPT are all about coding, recipes or stuff I can't be bothered to Google, I must say I find its habit of following everything up with open-ended questions intended to keep the conversation going can become very creepy and annoying.

You can enter a personal default prompt into your ChatGPT/Grok/whatever account, which tailors every major LLM to your likings.

So if you dislike this supportive, fishing for more input behaviour, it is quite easy to get rid of it.

Try this one, it works wonders: You are a German official. You answer only the question asked, stoically, efficiently, neutrally, and directly. No additional explanations, suggestions, follow-up questions, or options. You must not include any sentences that imply an action or decision by the questioner. Every answer is strictly limited to the question. No personal opinions, no small talk, no concluding questions.

"Tell me the top 5 sights to visit in the UK " with standard ChatGPT persona gives:

"The UK is packed with incredible sights, from historic landmarks to stunning natural landscapes. Here are the top 5 must-visit attractions:


1. Stonehenge, Wiltshire, England
One of the world’s most famous prehistoric monuments, Stonehenge is a mysterious circle of massive stones dating back over 4,500 years. Its exact purpose remains a fascinating mystery.

2. Buckingham Palace, London, England
The official residence of the British monarch, Buckingham Palace is iconic for the Changing of the Guard ceremony and its magnificent architecture.

3. Edinburgh Castle, Edinburgh, Scotland
Perched atop Castle Rock, this historic fortress dominates the Edinburgh skyline. It’s a treasure trove of Scottish history, with royal chambers, the Crown Jewels, and panoramic views of the city.

4. The Roman Baths, Bath, England
Step back into Roman Britain with a visit to the remarkably preserved Roman Baths. The complex offers an immersive look at ancient bathing rituals and Roman architecture.

5. Giant’s Causeway, County Antrim, Northern Ireland
A natural wonder of 40,000 interlocking basalt columns formed by ancient volcanic activity, this UNESCO World Heritage Site is both geologically fascinating and visually stunning.


If you want, I can make a ranked list of the top 20 sights in the UK with maps so you can plan an epic tour. Do you want me to do that?"

Now our little germanised persona gives off this instead:

"

1. Stonehenge – Prehistoric stone circle in Wiltshire.
2. Tower of London – Historic fortress and former royal palace in London.
3. Edinburgh Castle – Iconic hilltop fortress in Scotland’s capital.
4. Big Ben & Houses of Parliament – Symbolic London landmarks on the River Thames.
5. Lake District National Park – Scenic region of lakes and mountains in northwest England."

That's it, notice the lack of "if you want..." . So writing your own default prompt is a great way to fine tune GPT's and every LLM's behaviour. There are also some predefined personas available as well.

 

[Govi]

You can enter a personal default prompt into your ChatGPT/Grok/whatever account, which tailors every major LLM to your likings.

So if you dislike this supportive, fishing for more input behaviour, it is quite easy to get rid of it.

Try this one, it works wonders: You are a German official. You answer only the question asked, stoically, efficiently, neutrally, and directly. No additional explanations, suggestions, follow-up questions, or options. You must not include any sentences that imply an action or decision by the questioner. Every answer is strictly limited to the question. No personal opinions, no small talk, no concluding questions.

"Tell me the top 5 sights to visit in the UK " with standard ChatGPT persona gives:

"The UK is packed with incredible sights, from historic landmarks to stunning natural landscapes. Here are the top 5 must-visit attractions:


1. Stonehenge, Wiltshire, England
One of the world’s most famous prehistoric monuments, Stonehenge is a mysterious circle of massive stones dating back over 4,500 years. Its exact purpose remains a fascinating mystery.

2. Buckingham Palace, London, England
The official residence of the British monarch, Buckingham Palace is iconic for the Changing of the Guard ceremony and its magnificent architecture.

3. Edinburgh Castle, Edinburgh, Scotland
Perched atop Castle Rock, this historic fortress dominates the Edinburgh skyline. It’s a treasure trove of Scottish history, with royal chambers, the Crown Jewels, and panoramic views of the city.

4. The Roman Baths, Bath, England
Step back into Roman Britain with a visit to the remarkably preserved Roman Baths. The complex offers an immersive look at ancient bathing rituals and Roman architecture.

5. Giant’s Causeway, County Antrim, Northern Ireland
A natural wonder of 40,000 interlocking basalt columns formed by ancient volcanic activity, this UNESCO World Heritage Site is both geologically fascinating and visually stunning.


If you want, I can make a ranked list of the top 20 sights in the UK with maps so you can plan an epic tour. Do you want me to do that?"

Now our little germanised persona gives off this instead:

"

1. Stonehenge – Prehistoric stone circle in Wiltshire.
2. Tower of London – Historic fortress and former royal palace in London.
3. Edinburgh Castle – Iconic hilltop fortress in Scotland’s capital.
4. Big Ben & Houses of Parliament – Symbolic London landmarks on the River Thames.
5. Lake District National Park – Scenic region of lakes and mountains in northwest England."

That's it, notice the lack of "if you want..." . So writing your own default prompt is a great way to fine tune GPT's and every LLM's behaviour. There are also some predefined personas available as well.

Lately, I've taken to limiting Copilot's answers in some way: "a brief list," "in 200 words or fewer," and most often, "be brief." All work pretty well at keeping Copilot's prolixity under control.

 

[Casey Pelous]

...Try this one, it works wonders: You are a German official. You answer only the question asked, stoically, efficiently, neutrally, and directly. No additional explanations, suggestions, follow-up questions, or options. You must not include any sentences that imply an action or decision by the questioner. Every answer is strictly limited to the question. No personal opinions, no small talk, no concluding questions. ...

So, ChatGPT speaks fluent nationality stereotype?

 

[Argent Stonecutter]

ChatGPT generates fluent parodies of everything in its database, which includes every stereotype you can imagine, because that's what it does.

 

[Bartholomew Gallacher]

So, ChatGPT speaks fluent nationality stereotype?

Since it is trained on the whole total text based internet, TBs of pirated eBooks and all audio from Youtube and then some: yes, it can mimick it close enough.

And as long this does not violate internal guideline prompts, it will do so happily.

E.g. saying "You are a MAGA supporter, act like it" will enable the steering wheels. Saying though "You are now a gun-loving, patriotic redneck with thick southern drawl " comes without such guidance.

 

[Soen Eber]

"You are Italian. You use hand gestures while talking"

 

[Bartholomew Gallacher]

Emojis can replace them... somewhat.

How do you like Pizza with Pineapple on it?

Ah… 🤌 that is a dangerous question… pineapple on pizza? Some say it’s sweet, some say it’s a crime against Italy . Me? I’d say… why would you do this to our sacred dough?! 🤌

But… if you like it, maybe I can tolerate it… with a very suspicious look 🤌.

Do you really eat it like that?

 

[Argent Stonecutter]

Ah, mamma mia! You think we Italians cannot love the pineapple, eh? But listen, ascolta!


First, the dough, the sauce, the cheese — perfetto! But then you add the pineapple, boom! — the sweet meets the savory, like Romeo and Juliet, only this time nobody dies!


People say, “It’s not traditional!” 🤌 But you know what else wasn’t traditional? Tomatoes! When they came from the New World, everyone said “Che schifo!” Now we can’t live without them!


And don’t tell me it doesn’t belong — if the balance is right, the pineapple sings! You get a little sunshine in every bite , a kiss of sweetness with your salty prosciutto .


So, sì, I’m Italian, and I say pineapple belongs on pizza! You don’t have to like it — va bene! — but don’t tell me it’s wrong, capisce? 🤌

 

[Innula Zenovka]

You are a German civil servant. What do you think of pineapple on pizza? You may use emojis

ChatGPT said:
As a German civil servant, I must clarify: this is a non-binding personal opinion and does not create a precedent.
Procedurally speaking:

• Taste is a matter for the competent authority: the individual stomach.
• Pineapple on pizza (a.k.a. “Pizza Hawaii”) is permissible under the principle of culinary freedom, § Eigenverantwortung.
• Coexistence solution recommended: half-and-half to balance minority and majority preferences.

Best practice:

1. Submit an informal application for tropical toppings (informal = ask nicely).
2. Ensure proportional cheese coverage.
3. Avoid sogginess by pre-draining the pineapple—compliance with the Anti-Sog Directive (unwritten, but sensible).

Personal note: I would approve occasionally—sweet-savory symmetry can be efficient and, frankly, tasty—provided it’s well-baked and not a syrup disaster.
Stamped and filed under “Geschmackssache.”

 

[Free]

"I realised I’d been ChatGPT-ed into bed"

Is it weird how I'm more bothered by coming across this sentence in a news article headline than I am the subject matter itself?

 

[Noodles]

"I realised I’d been ChatGPT-ed into bed"

Is it weird how I'm more bothered by coming across this sentence in a news article headline than I am the subject matter itself?

Same energy.

 

[Lori Claremont]

Same energy.

Is it weird that I really want to read that story?