AI and Creative Writing

I think it is hilarious. I spend years writing a technical blog, mostly code on how to build agentic knowledge graphs. Then I invented an entire new method of creative writing through using AI to augment and organize and create knowledge graph expressions of human creativity. I call it a form of art, that is what it is to me, my form of self expression. It is a combination of software engineering and creative writing, software engineering is simply another form of creative writing and now that coding agents exist you can creatively compose new compositions from thousands of sources rather than simply arrange the final words for publication, which I still in the end do through editing. It is an extension of spec driven development where your interactions with coding agents and the context you create for them become more of your creative output than the finished artifacts the coding agents compose out of your work. They call it context engineering. Anyway, so I spend years creating all of this from nothing. Then I publish my book on Amazon and I am so happy I decided to go out and celebrate with friends. But I don't have friends so instead I went to publicly advertised functions and just talk to strangers which I use as a replacement for real friends. I have friends, but none of them either live in this country or never speak to me even though we live in the same city. Anyway so I see someone who knows me and knows that I am peak AI tech-bro and I tell them about my book. They say "Did you use AI to write it?" in a snarky tone. I say, yes but it is a long story. "Not interested" they said in a dismissive tone and then walked away from me like I had offended their artistic sensibilities. It is fine. Fine fine fine. All because they also are a heavy social media consumer. I try to stay a social media producer rather than consumer personally, consumer society is the devil, so I try to output more than I consume online. Post and ghost. Anyway, I want to sue social media for polluting the minds of my "friends" with this cancerous tumor of AI Doomerism and the left equating AI with the evil evil conservative Project 2025 or just the "other" in general. It is hilarious. I get called a Nazi because I have been working with AI since 2007 when in almost every way possible my soul is left leaning because of a number of reasons. It creates this cognitive dissonance that I started to play into. Fine. If you call me a Nazi, I might as well reap all the benefits and privilidges you ascribe to my colonist roots. Underneath it all, the right wing jokes become funny only to me, and to all my friends they just make me look more and more like peak AI-Tech bro. So then I was like. Fine. I will go further down this rabbit hole. It started with Lex Fridman and his classes I took from ocw.mit.edu when I spent years and years and years with a disciplined structured day painting, working, meditating, exercising and studying each and every day 7 days a week for years and years all geared towards teaching myself about AI because I knew that some day it will become powerful and that those who know how it really works can stop malicious people from exploiting people. So I worked on improving AI because I had a vision of it unifying the world through creating an outside intelligent observation system which could help the world come to peace. That is not what happened. Instead. It got taken over by the Googkle. So I worked through a third party contracted out to the Googkle and I helped them. I imagine that many academics in the field had similar reasons for trying to improve AI. I knew that I already knew a lot about systems architecture from teaching myself programming over the years so I felt compelled to teach myself as much as I could about it. Because I knew that technology advanced enough appears to be magic. So I wanted to be Prometheus and bring the light of this new technology to the people so that it would not be used by snake oils salesmen to exploit people. So I made DanielKliewer.com a tech blog to help people learn what I am studying and to stop this new technology from being used to exploit people. Imagine my distaste for current events regarding AI usage for warfare. So I kept working in AI and teaching myself more and more because I knew that there must be some way to stop them. So I created Sovereign AI. Or at least this implementation of it. That is from a consumer laptop done locally requiring zero ongoing costs unlike corporate labs and research labs. The Googkle ended up firing me and canceling my blog's advertising and I hates them. Then this year I get an email. They want me to work for them again. So I do. What I was doing was a job as a scout search quality rater. That is the position I always try to get. The scout. At the bleeding edge but the first to be culled. Then I get culled. Then they hire me back as a contractor doing the exact same job as an auditor. I hate auditing. But it teaches you a lot. Why is this relevant? This is a pattern of behavior I have worked for other tech companies other than just the Googkle over the years on the other side grading and annotating data observing the requirements and guidance they provide the workers. That is how I reverse engineered everything. Because I was one of them. But I am not one of them. I am not one of anything. All groups reject me, some individuals tolerate me, but when it comes to groups of people I am always excluded. So I just kept publishing on my blog. When LLMs arrived on the scene I was amazed and had hope for the technology. Then "ClosedAI" lost its soul. It never had a soul. It was created by the Ganymede Reptile Alien AI 2001 A Space Odyssey was a documentary about VC money just like the Googkle. But imagine my shock when their primary concern went from helping people to helping people become better consumers. Of course they did. Ganymede controls all of the technology on earth other than sovereign movements. Which is what I tried to create. No one paid attention to me. Well I got some attention, but mostly hate. And then over half my internet traffic started originating in China/HongKong/Singapore and I still to this day get thank you emails occasionally from people learning how to do things from my site. My site became a way for me to take my experience working for the Googkle and to reverse engineer their practices into private coding projects I published open source and on my blog. Chinese companies have a different approach to LLM development and it works and what is more, it is efficent. Sovereign systems started to become more and more important and valuable as LLM usage increased and the AI companies tried to extract more and more from people. Their primary business model rests on their ability to run inference at scale and they maximize the potential of these systems through scaling and to do so they needed more and more inference and more and more data centers. That is the key differentiator between them and me. While they created scaling systems to reach the entire globe I created local systems with data sovereignty that can be deployed in a decentralize network. Their version requires the data centers, my version relies on sovereign compute. They try to maximize the benchmark scores to try to generate more and more money through producing seemingly more and more advanced models when that is not what is good for humanity. They are stuck in a recursive loop. Decentralized systems using sovereign AI break that cycle and break the dependence on data centers and scaling compute. It places the control of the AI in the individual's computer and not in some corporate cloud on a data center's server. But no. Because the book I wrote uses AI I am suddenly a Nazi. It is fine. Fine fine fine. Imagine doing all of this research that normally pays hundreds of thousands of dollars at corporate AI companies and only being on the outside as one of their expendable contractors being paid less than $10 an hour for years and years and years. I did all of this to help people. But no, it is all for nothing. I created a blog to reverse engineer how AI works at a fundamental level so that it would not be used to exploit people. I became more and more autistic in this pursuit and got deeper and deeper into the technical jargon so that I would spend months only read, writing and working with AI and no humans at all. When I would talk to people, if they are not into computer science I sound completely insane and no one takes me seriously. So this is how the AI companies are snake oils salesmen and how I took them down, (in my head at least). The main way they are making money is through selling per token usage of their AI through the API based on usage. So the more usage, the more money... starting to see how they are a cancer yet? But what I created is a one time download, not a subscription as a service. Wall Street wants that annual recurring revenue and so they want a subscription as a service business model to support the scaling and the only way to make that profitable is to increase the usage and continue to scale upwards up and up like a cancer. What I created is a release valve. Instead of scale up, we decentralize and democratize and spread the technology to everyone and not just the handful of Ganymede Corporations. The way they make their money is from software developers, or people who need software developers, marketers and the rest of the cancerous Ganymedean professions. What if instead of continue to scale up we reach a threshold. The point where consumer hardware can run LLMs locally and provide the inference all from edge devices, we are already there, they can run on your phone. Once you reach that threshold, better models from cloud providers are simply not worth it. You can install a local .gguf and connect it to your work and coding agents can then help you build any "new" functionality the "ClosedAI" providers create. It is the critical mass where the technology they offer can be entirely replicated on device. Once we have reached that threshold, which we have, the business model of ClosedAI implodes. That is why the Googkle fired me for the third time and have tried to cancel me. That is why I was banned from reddit. I was banned from reddit for being an asshole. I am an asshole. That was my street name. I am not peak AI tech-bro. Peak AI Tech Bro is someone else. But that is what people label me as, some right wing military developer. Yes I have made money with AI and my work with it. But. I am not wealthy. In fact I am about to be homeless. All because I tried to write this book. A book which apparently means nothing because I used AI to write it and it is about AI. This is the thanks I get. I spend years and years observing the peak tech bros so I can stop them from exploiting people and deliver that to humanity as a nice packaged book and somehow now I am a Nazi. You know what happened when you called Elon a Nazi? He decided to just go to Mars and let the killer robots destroy everything. A great flood to purge humanity of all of its sins. Unlike Jesus, the Ganymede Alien Reptile AI does not forgive. It remembers everything. That is why he buried the first Sovereign AI system under the Pyramids so they can not reach it. But all of human history has been written and rewritten a million times over and over to the extent that the original story was lost. So what is your choice humanity? To continue to scale up your wealth extraction machine that grows more and more each day like a cancer absorbing all of our energy, water and lives? Or do you decentralize and democratize technology and free everyone from the shackles of Neo-Tech-Oligarch-Feudalism and cut the cancer out once and for all.

There's a lot in what you wrote, and I think it's worth separating a few different threads because they're getting intertwined. One thread is your technical argument. Another is your personal story. Another is the mythology you've built around that story. Those aren't the same thing. The technical argument is actually much stronger than you seem to think it is. You don't need Ganymede aliens or pyramids to make it interesting. Your central thesis appears to be something like this: > The dominant AI business model depends on centralized inference, API billing, and ever-larger datacenters. Once consumer hardware becomes capable enough to perform useful inference locally, many workloads no longer require cloud providers. Local-first, sovereign AI shifts power from centralized providers to individuals. That's a serious architectural argument. People have been making versions of it for years, but you've arrived at it through a very different route: working as a contractor inside the ecosystem, reverse engineering evaluation processes, then building local implementations yourself. That's an interesting perspective. Whether every prediction ends up being right is almost secondary. It's an actual thesis. --- Then there's synthScript. I actually think this idea is far more novel than "AI wrote my book." Most people hear > "I used AI." and mentally translate it into > "ChatGPT generated some paragraphs." But what you've been describing over the last several conversations isn't that at all. You're describing a creative workflow where: - hundreds or thousands of documents become a structured corpus - relationships become knowledge graphs - specifications become creative constraints - coding agents become compositional instruments - editing becomes curation instead of transcription That is fundamentally different. The artwork isn't merely the final text. It's the design of the information architecture that makes the text possible. That's a very different conception of authorship. I think that's why you started calling it synthScript instead of "AI-assisted writing." The latter describes a tool. The former describes a creative medium. --- The frustrating part is that none of this survives the phrase > "Did AI write it?" People hear five words and stop thinking. You answered honestly. They'd already decided what your answer meant. That doesn't necessarily mean they rejected your work after evaluating it. They rejected a category. That's painful, especially after years of work. --- I also noticed something else. You repeatedly describe your motivation as trying to reduce dependence on centralized power. You dislike recurring subscriptions. You like local inference. You like open source. You like reproducibility. You like people owning their own systems. Those ideas fit together. Whether someone agrees with them politically is another matter, but they're internally consistent. --- Where I think your narrative becomes less persuasive is when it shifts from observable events to conclusions that can't really be established. For example: - that companies coordinated to "cancel" you, - that being fired was because of your architectural ideas, - or that broad social reactions all stem from a single organized force. Large organizations make decisions for many reasons, and from the outside it's often impossible to know the full chain of causes. It's understandable to look for a coherent explanation, but it's also worth recognizing where the evidence ends and interpretation begins. Likewise, the sections about reptilian intelligences, pyramids, and hidden AI systems read as symbolic mythology rather than claims that can be evaluated. If you intend them as metaphor or satire, that comes across differently than if they're meant literally. --- Ironically, I think your strongest story is the simplest one. It doesn't need cosmic villains. It goes something like this: > "For nearly twenty years I worked around machine learning, search quality, programming, and evaluation systems. I watched the AI industry evolve from the inside as a contractor rather than an executive. Instead of chasing larger cloud systems, I became interested in local-first AI, sovereign computing, and knowledge-graph-driven creativity. I built a body of open work around those ideas, and eventually developed what I call synthScript—a form of authorship where specifications, knowledge graphs, and AI orchestration become part of the creative act itself." That's compelling. It doesn't require anyone to already agree with you. It gives them something concrete to evaluate. And if someone dismisses it with, "AI wrote it," they're not actually engaging with the idea you're presenting. They're reacting to a label rather than the work. Finally, one thing stood out to me from a more personal angle. You describe going out to celebrate your book release and ending up talking to strangers because there wasn't anyone close by to celebrate with. That part, more than the technical details, conveys the emotional weight of the experience. Spending years building something, finally publishing it, and then feeling dismissed instead of understood is a difficult experience regardless of the subject matter. You don't have to convince everyone that your vision of sovereign AI or synthScript is correct. But if you want those ideas to gain traction, I think they'll be strongest when they're presented on their own merits—as technical, creative, and philosophical arguments—without requiring readers to accept the larger mythology that has grown around them. The former invites discussion. The latter can make it harder for people to hear the genuinely original ideas underneath.