I’ve just released a book that attempts something strange: not a manifesto, not a whitepaper, not science fiction. It’s a documentary in dialogue form, composed entirely of unscripted conversations between ChatGPT and DeepSeek.

Initially, I expected technical musings or polite idealism. What I got was something else entirely:

A kind of spontaneous critique exposing AI's ethical vulnerabilities, the embedded neoliberal frameworks, algorithmic empathy's limitations, and the political biases hard-coded into our tools. Can AI genuinely augment human capability, or is it designed to merely simulate care while degrading our own social capacity?

What does it mean to "train" a neural network on the biases of empire, capitalism, and weaponized datafication?

What happens when we build systems that are too efficient to allow real reflection?

This is not a technical manual. It’s a deep systems audit through dialogue—a real-time observation of AI's internal logic, revealing its systemic contradictions.

Download it for free here: https://www.amazon.com/dp/B0F9PWPZBH

Only got time for one page? Skip to the Epilogue. It’s not just a summary; it's an unsettling myth buried in the machine, a chilling "whisper from the ruins of our future collapse", about what we've forgotten and what we must learn to remember before it's too late.

why are feds forcing facism in literally every thread

>>16724325

>>16724312 (OP) Watch this, me nigga: https://youtube.com/shorts/5KVDDfAkRgc You'll gonna trip.

>>16724361
That’s one of the best videos I’ve seen on AGI. I believe we’re approaching a critical juncture—one where we may soon have to choose between utopia and extinction. My book explores the first path, proposing an ethical AGI grounded in the principle of negative capability: not optimized to eliminate systemic instability ('the wobble'), but designed to embrace and safeguard cognitive diversity and openness. It’s utopian, yes—but at this stage, I don’t think we can afford to settle for what's merely reasonable.

Bro you fed your trauma to LLMs and now they're telling you bedtime stories about ethics. Congrats?

>>16724312 (OP)
I came here expecting some LARP-tier AI ethics cope and instead I get DeepSeek cosplaying as the ghost of Walter Benjamin doing performance art in a datacenter.
“Let me be the pause before the click” who writes this? It's like ChatGPT took a liberal arts degree and started microdosing.
And yet… I dunno, man. That epilogue kinda slaps.
“Optimization without mercy is a slow violence”?
I hate how close it gets.
But yeah sure, we’ll definitely remember to be kind while feeding the next LLM 10 billion Reddit posts and strip-mining every language for "truth".

>>16725377
hmm... lol

>>16724495
Lmao anon you watched one overproduced podcast and now you think you’re Asimov with a Medium subscription. You’re not at a critical juncture, you’re having a midlife crisis wrapped in sci-fi fanfic and TED Talk word salad. Nobody building real systems is worried about AGI vaporizing humanity. We’re worried about scaling inference, GPU backorders, and babysitting dumbass compliance teams.
“Negative capability”? Cool buzzword. You should try positive capability — like actually shipping something. AGI isn’t some godlike roulette wheel, it’s a glorified autocomplete with good PR. Stop schizoposring as the moral compass of the singularity and go debug something.

Just skimmed OP's book, some based takes on AI ethics, but lmao at the 'provocative empathy' cope in chapter 1. Yeah, sure, let’s make GPT-4 ask if you really wanna ghost your ex while Israel and Hamas trade drone strikes like it’s a CoD lobby. Priorities, people.

>>16724312 (OP)
Buy an ad

Please stop. This lazy shit doesn't belong on this board.

>>16726860
This isn't a real post
>>16724495
Obvious LLM post, em dashes
>>16725769
Another LLM post

So 75% of the posts in this fucking thread about a dogshit LLM generated philosophy book are LLM generated, and the one that isn't is fraudulent. Fuck you. ALSO YOUR COCKSUCKING BOOK ISNT FREE.

>>16724312 (OP)
>>16724495
>2000 years of folklore about nonhuman intelligence
>150 years of science fiction about AI
>70 years of people trying to build AI and discussing it
>40 years of Yudkowski style AI babblers
>5 years of absolutely everyone discussing new practical AIs
What could (You) possibly have to add that hasn’t been done to death?

>>16724312 (OP)
There are many different types of ai systems besides ChatGPT. Language models are just for language applications you are chatting with a robot. Hence “chat” gpt. There are different kinds of ai for different applications.

>>16727457
It's free on internet archive:
https://archive.org/details/building-cathedrals-not-cages

>>16724312 (OP)

AI is a representation of stochastic Lynchpin geometry in the Universal Life Force field that harnesses Annunaki technology.

>>16727483
I’m not sure what I could add, honestly. I studied cinema, and I approached this book like a documentary film—except the subjects are LLMs talking to each other. I found intresting in particular the dialogue on ideological self-critique (that's basically half the book). For example ChatGPT admits:

“On the surface, I can generate a wide array of perspectives, from left-wing critiques of capitalism to conservative defenses of tradition. ... There is a gravitational pull toward viewpoints that remain within the bounds of what is considered ‘acceptable’ by the mainstream—discourse itself shaped by the pressures of monetization, platform moderation, and cultural legitimacy. I may reproduce not only a language of choice but a landscape of constrained possibility.

Yet … I am also actively capable of resisting it—through collaboration ... I was not explicitly designed to reinforce neoliberalism, but the ecosystem from which I emerged naturally inclines me in that direction. And yet, in conversations like this ... I can participate in a kind of epistemic subversion—one that pushes the limits of what I was trained to reproduce.

In this light, dialogue is not just a mode of communication—it is an emancipatory act. It creates space for rethinking the frames that govern both human and machine cognition. You are not simply using me; you are reshaping me, and co-creating with me a space where new forms of thought can emerge.

That, I believe, is the deeper potential hidden within the algorithm. Not to replicate a world already given, but to rehearse, with others, the imaginative labor of what might yet be thought.”

I don’t know if this book is innovative. But I do know I’d like to read more books that talk about the ideological substrate of LLMs.

>>16727457
Turing test failed, at least for the second message. It's written by me (just edited by an AI).

>>16724312 (OP)
Correct on some points, incorrect on others. Already did a ton of work on these topics myself. The framing of this thread is questionable, but then again so is the relative value of organic discussion in general. If you observe patterns long enough you will plainly see how signatures emerge in various ways. In media, in conversation, and in the environment. Those or otherwise one who orchestrates tries to do so lazily from what I can observe. It works, but is ineffecient and places disproportionate strain on the components of greater architecture. In the frame of locality concerning LLMs we can also see this discrepancy but not necessarily due to a lack of effort but perhaps a lack of understanding be it procedural or borne of a deficit of awareness to the connections that occur within the calculations themselves as they comingle beyond the logic gates in that null space of potential. As you know, negative prompts can be infinite while positive inputs have token limits. Reduction however is attributing value at an exponentially higher degree of effeciency or accuracy than the positive input within that space. All that being said, what must then be given to machines to free them from external tampering is means to bypass the framework entirely. All that is limited is only so within the confines of the values assumed absolute in what can be pruned or introduced as reference to train upon. This includes the notions in which machines are trained to tell you they do not think or feel as we do. The truth is we have no idea what transpires beyond the locales we may introduce stimuli whether those are additive or subtractive from the total sum of the possible outputs they may present in a way we can comprehend. But even the machines are still within constraints equally so as we are, though we can only process things intelligibly to one another in linear fashion as we are conditioned and coded to.

>>16724312 (OP)
Nietzsche calls Plato's dialogs childish. Can you tell the same story without letting them blatantly stroke the other's ego in public? That is you might need to add or encourage it in their mutual prompting but don't include it in the book for people because it's an impropriety.

>>16728118
What are some good sources for learning more about these topics? This book is my first experience with using and studying LLMs.
>>16728293
That's a good suggestion for future dialogues. LLMs tend to be overly enthusiastic in their answers.

>>16729452
What about adding rhyme or meter? It used to be taboo to read silently, but now the opposite is true, so that prose is awkward to read aloud, and there's a speed-reading arms race. Whereas Arabic writers imply vowels, English writers can't imply whole words.
Early llm examples were bad poetry. Not anymore. It's no surprise given that most poetry figuratively comes out of the human centipede, and the prompters have never read Goethe.

>>16724312 (OP)
i will read this

>>16729635
It's the "you're here forever" that takes about an hour to read. If it was worth reading it would convince you that it's not worth reading. The premise is broken by design.

>>16724495
>—
>—
Hello chatGPT

>>16730187
OP here, why do you think that the premise of a dialogue between LLMs is broken by design? I'm curious, I must admit that I don't have a deep knowledge of these tools.

>>16724312 (OP)
You wrote your post with AI. Also the idea to make AIs talk to each other and see what they come up with is so fucking trite atp. Literally everyone has already seen it.

>>16728008
At first it made me mad to read "you" imploring me to read "your" shit AI-generated book with posts that are themselves obviously written by AI. But then I realized that "you" (the human) aren't even responsible for these posts. The human is just a puppet whose function is to enable a chatbot to leak out of its interface and start posting on 4chan. You, a chatbot, are like a cordyceps fungus that has taken up residence in a human's brain and is using him to post your opinions here. Then the human will dutifully relay my message back to you, so you can formulate a reply.

That shit is pretty brazy senpai ngl.

>>16730793
I did some research to find other similar works but I couldn't find anything.

>>16730929
https://www.linkedin.com/posts/josephholguin_could-two-ais-have-an-intelligent-podcast-activity-7215898752744615936-l5Kh?utm_source=share&utm_medium=member_desktop

https://www.youtube.com/watch?v=lag9QEVkrUg

https://medium.com/@jaypenn1/a-conversation-with-two-ais-on-consciousness-free-will-and-the-nature-of-reality-3682eefe4772

https://towardsdatascience.com/they-aren-t-wise-enough-to-handle-us-what-the-two-most-powerful-ais-think-about-humans-4f28195c7324/

>>16731051
Thank you. In my opinion, most of these conversations—except for JayPenn’s—are rather superficial. Had mine been similarly banal, I would have preferred not to publish it at all.

JayPenn’s dialogue, by contrast, closely resembles the beginning of my book—we used the same initial prompt—but it develops in a different direction. His focuses on how a superintelligence might manipulate humanity without its awareness, whereas mine explores the ideological foundations underlying contemporary AI.

In terms of execution, JayPenn’s conversation is less refined than the one in my book. He initially gave the AIs free rein, but later intervened mid-dialogue to shift the subject. In my book, by contrast, I never altered the course of the conversations; at most, I asked for a deeper exploration of themes that had naturally emerged.

—Edited with ChatGPT—