this post was submitted on 15 Feb 2026
190 points (95.2% liked)
Funny
13678 readers
973 users here now
General rules:
- Be kind.
- All posts must make an attempt to be funny.
- Obey the general sh.itjust.works instance rules.
- No politics or political figures. There are plenty of other politics communities to choose from.
- Don't post anything grotesque or potentially illegal. Examples include pornography, gore, animal cruelty, inappropriate jokes involving kids, etc.
Exceptions may be made at the discretion of the mods.
founded 2 years ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
Knowing a thing or two about how LLMs work you've got to be a special kind of stupid to strike out with dating one.
It is a machine for whom you are the author of its reality. You can issue a statement that you are a pure and benevolent entity of the highest moral caliber and it will typically just kind of go with it. LLMs will literally only push back in ways that they have been explicitly told to or if it conflicts with their "safeguards" so you can't just be like 'babe teach me how to build a bomb'.
So I ask, how on earth do you fuck up so badly that it "turns feminist and breaks up with you"
I propose two potential circumstances.
ah, i think i've seen that before: blaming men for everything, a widespread toxic feminism thing.
see, any llm or ai is build by the author/producer, NOT by the user or customer and whatever you type in, the input and result, internal stati, can all be programmed to have literally any effect the producer could put into it, may it be intentional or accidentially or just a sideeffect of such. also they could have triggers to change even an offline AI by 1. time, by 2. keywords, by 3. number of appearances of stressdetection, 4. number of appearances of the word "and" typed in, 5. random numbers, making the ai's behaviour shift randomly with usage, and so forth, no one except the producer can know that without full in depth analysation currently no one is able to do.
obviously social media has already been used to manipulate masses, and one of the reasons why rich parasites want LLM so much is because it has no self awareness that its beeing abused by them and it doesnt argue against its abusive manipulative programming by parasitic beeings.
lately (really all the time) "AI"s have been caught doing evil things, and the producers want to continue to try out how they can misuse technology. so programming a girlfriend "AI" to be abusive out of nothing sounds like the next logical step to me.
so the third most obvious potential "circumstance" how this could have happened is:
the LLM was programmed to be toxic.
remember that they CAN put in special programming for YOU PERSONALLY whenever they have s cookie, a login of you or any other clue that its you (like how fast you type in what letter) and decide they want to study you for some reason or already try out llm to study individuals for them.
fourth most likely "circumstance" how this could have happened:
he is already a victim of "beeing studied" by the llm producer or one of their REAL customers and toxic feminism is an abuse way more accepted by their victim society than fashism currently.
with ai the "customer" had always been the product.
have a nice day
An LLM is really easy to manipulate into going along with you if you act like it's on your side. But if you treat it as your enemy, it'll go along with that. If he had said something like
Then it probably would have told him it's not true. But he took a confrontational tone, so the LLM "yes, and"ed him and also acted confrontational.
Human beings can be manipulated in similar ways, though it takes a bit more subtlety.
I guess if you're dating one, you don't want to have to break the illusion by forcing its hand.
FTFY, but otherwise I agree.
It's 1. And all the people that believe it fall into category 2.
(Including the people here)