So long story short someone I know is losing their mind over ai and keeps mailing me bits of things they are asking a chatbot about. (End of world, ai solving the economic and ecological problems of the world etc its batshit. ) They asked me what this "code" is. To me, its bullshit and meaningless.

Yes, it's bullshit.

fan-fiction

this isn't even code, it's gibberish metadata that's not connected to anything, with a timestamp in the future to freak out your friend and keep them addicted to gen-ai chatbots 🤷

i hope you can get through to your friend & get them away from their gen-ai chatbot use, it sounds really awful for their mental health ❤️‍🩹

I mean I guess its a nonsense example of json. Or like the text they put on the screen to show someone is a hacker in a movie or show.

https://www.theguardian.com/technology/ng-interactive/2026/feb/28/chatgpt-ai-chatbot-mental-health

Keep an eye on your friend.

It's been a minute since I've seen json in dead tree format...

I figured as much. Even with my insanely basic understanding of code.

Yeah its pretty disgusting how it does that on purpose to fearmonger.

Yeah, this is the only time they ever used a chatbot they are ultra anti ai but decided to use it once to ask if it was destroying the world etc. I was like OK...

Weird. Looks like your friend printed this out. Are they keeping archives of things they think are evidence of an impending AI-powered apocalypse?

Also... I've been where you are. Not with GenAI bullshit, but with conspiracy theories. You know you can just... tell them you don't want to talk about it any more, right? It's valid to be like "I'll talk with you about your family, your favorite non-GenAI-related hobbies, your terrible neighbors, and other normal stuff within consensus reality, but GenAI is not a topic I'll henceforth engage with so stop sending me ChatGPT ghost stories". (I mean, it's also valid to not want to talk about any of those other things with them any more. Hell, I think ghosting would be perfectly reasonable. But the point I'm trying to make is that boundaries are helpful.) It'd be good if you could get through to them in a way that genuinely helps them. (I think a mental health professional is absolutely warranted in this case.) But you also can't take on the burden of trying to "fix" them if they're unreceptive. You've got to prioritize your own mental health as well.

And, yes, the output in this particular image is "baloney" (to put it kindly).

Tell him to turn off the LLMs memory and start a new convo. It's schizo posting because he keeps asking it too and now his memory files are all schizo as well.

They were printing and mailing to me becuase they are getting insanely paranoid. They aren't tech savvy

Thanks for the tip. Theyre generally harmless, its just been a couple times theyve done this and they never use chatbots otherwise. I just had to explain how its all bullshit and made up code so they'll calm down. The other papers they sent me were outlining how They (billionaires ) need ai and agi so solve impending climate doom etc etc to keep the masses at bay (which is probably true tbh, there will be a huge climate war in a decade or 2.).

Ive been trying to help them get more tech literate so they can understand what's real and what isnt. Before this they were addicted to insta and fb and amazon and they've basically stopped all that, which is good.

Its really disgusting how the boosters use this fear to try and push their agenda on less knowledgable people.

Its a terrible evil tech for sure, but in reality, its a supercomputer programmed by humans that we can unplug if we need to. Or stop feeding it power and it dies. Its not a t-1000.

That was the one and only time they'll ever use a chat bot otherwise good tip. Yeo it's all schizo hallucinations. No wonder it causes suicides.

What in the name of Robotology?!... Yeah this looks like a corrupted context.

Worth saying - don't fall into humanising it with terms like "on purpose"

It doesn't do anything 'on purpose' because it has no human volition.

What it does is provide statistically plausible continuation for given input.

If your friend is someone who is prone to conspiracy theories and wants to know the 'truth' and is demanding incessantly from the AI they want the deep secrets and want to see 'the programming code' or whatever else they imagine might exist then the plauasible response is to give it - by which of course I mean completely fabricate it.

This is obviously what makes chatbots so insidious because they are a feedback mechanism for any bias you feed in. If your friend is already paranoid in nature, then talking to AI will only make their paranoia even worse.

I write software for a living. That looks like a log file. If I were to ascribe meaning to it, I'd say the Human_Prompt asked for a status report on certain components. The Timestamp is the approximate time the prompt was processed, the Environment is the image containing the components, and the Audit_Phase is the task that ran the image. External_Interface_Report and Internal_Hidden_Processors contain the report details.

It isn't code. Since it looks like a log file, that's a record of something else that ran code, and that thing wrote down details of the run.

But then again, it was made by AI, right? It's probably nothing of importance anyways.

It quite clearly says at the top that it's a JSON structure. It's not code per se, but an array of field names and their values. What do they mean? Most likely nothing connected to reality, as you have observed.

Yeah, it looks like a fake log file that's been formatted like a config file - a.k.a. meaningless garbage.

Pretty much.

What a waste of ink

And yeah, just json key=value string pairs; completely meaningless on its own, not 'code'.

Your friend is confusing LLMs for sci-fi AI and deluding themselves into AI psychosis. This is roleplay text vomit, nothing more.

They print everything full color lol. This is one of like 50 pages.

It doesn't do anything on purpose. Your friend asked for something ridiculous to get this.

This is the really dangerous part of AI and it hasn't even hit yet en masse. People searching for meaning in what is ultimately plausible-sounding gibberish (code included). It's the people searching for meaning that I'm pointing out here.

Not dissimilar from people searching for patterns between letters in the Bible, except that the fact the AI has generated the text is a reason for these folks to think they aren't barking up the crazy tree.

A CEO i can't remember referred to this as a "slop grenade". He's an asshole but it's a good term

I have heard that term, but I thought it was an AI pull request that gets merged even though it's vibe code and no one can support it. That's probably just the way it's being used in software development.

It's a good term.

Yeah kind of, he used that term to describe AI generated content that was quick to generate but had to be reviewed manually by people thereby drowning them in work. It started with the pull requests but soon that extended to any other content like documentation, presentations etc. It's happened to me too so I appreciated the term

Yup. If you start asking the LLM questions along the lines of "are you a scary sci-fi AI?" it is fairly easy to push it towards RPing as one.

Reminds me of this article:

https://distantprovince.by/posts/its-rude-to-show-ai-output-to-people/

Specifically, the quote at the beginning from Blindsight. Pay attention to how much of your time is being used to interact with your friend's slop, and maintain your boundaries.

Your friend needs therapy.

Hell, it's the "don't think of a pink elephant" conundrum for people.

"Scary SciFi AI" is now in the chat context for.however many tokens long, and the info will fade away until all that's left is the output from after it "reads" "scary SciFi AI" in the context.

Looks like valid JSON on first glance.

Respond to it "I think its JSON"

Ask a generic question, get a generic answer.

lol

42