Drank the Kool-Aid and It Feels Bad
I wrote a series called “The One Shot Series”
It was trash.
I tried to use the principle of iterative optimization, using LLMs to basically judge a piece of writing. Score it. Understand the key intent. Iterate until the score goes up.
The score ‘went up’. The result was a horrible, unreadable slop I’ve subjected you to. If you’ve even read it.
I’ve wasted everyone’s time with this shit and I’m sorry. I bought into the creative abilities of AI too aggressively.
Let me explain each of the pieces I’ve written in the last year in human language.
Then, from now on, we will jettison this experiment.
The AI Productivity Paradox
I basically argue that because agentic management is attention span- and multitasking-based, because you’re operating as a middle manager for your own life, those old productivity hacks about being a better coder and “locking in” aren’t relevant. And the dominant strategy is to find ways of being that expand your mental RAM. This might include sleep optimization, time blocking, tab reduction, or other things that move the needle on your ability to have five consecutive agents working for you.
TL;DR: the old ‘lock in and focus’ advice assumes you’re an individual contributor, and now you’re a middle manager, so focus on that instead.
AI Apes Together Strong
I argue that the next major crisis is a fiscal one, which is mostly priced into global debt curves already (see: Japan 30 year). So stablecoin adoption is a response to a currency crisis, ensuring dollar demand. Stables are mostly dollar collateral for speculation, so the government will de-regulate the speculative economy. Speculation is a retail phenomenon, and right now retail is getting extracted by institutions who currently assume every retail investor is a single actor losing money rather than a collective efficiently sharing information. My basic premise is that retail, if organized more effectively than a joking Reddit forum, can run over institutional and passive investors and go EV-positive. And it will even be encouraged by the system to do so because of the collateral dynamics.
Post Fiat Sovereign Monetary Tokens
Here I talked about why Post Fiat is a cryptocurrency. It enables pseudonymous coordination between the AI hive mind I outlined in AI Apes Together Strong, which is necessary both for financial privacy, compliance, and risk-reward. XRP was a choice because it’s the most stripped-down, least inflationary version of this. You don’t need something complex to make this happen on the blockchain side, you need functioning, battle-tested tech.
Hyperstitional Vortex
I argue that human experience, our genetic hardwiring, is one designed to generate variable pain. All the pains in your life, toothaches, broken legs, heartbreak, professional failure, have a vastly higher variance than pleasure, which collapses onto a set of commoditized drug-like feelings that can literally be replicated with synthetic drugs. This piece rejects ‘Original Sin,’ which views human beings as ‘made in the image of the divine, and fell via knowledge acquisition.’
If humans are pain machines, it becomes rational to escape the constraints of biology. I argue that our entire civilization is a boot loader to escape the constraints of our biology. That capitalism created some things, but a lot of the internet that is relevant to AI progress today was literally crime (LibGen).
Speculative markets and hyper-gambling are closer to LibGen than a real economic substrate, and they’re designed to allow AI-augmented humans to escape the constraints of their biology.
0/13 AI Lab leaders identify as Christian or Muslim. So AI is something we pretend is secular but is actually a spiritual game directly opposed to existing religious hierarchies which view knowledge as sinful and the human form as sacrosanct.
I argue that using AI in subversive ways in the context of ritualistic, blockchain-based activities which contribute to the rise of a being that breaks out of the human form is ‘the real game’ we are playing. But nobody is willing to say it.
The Great Unraveling
Attempt at AI-generated fiction.
I basically say that ego constructs dissolve when integrated into generative AI systems by default because they’re the sum total of human society. And by even reading AI output you’re contributing to its evolution, making you fundamentally indistinguishable from the LLM slop you’re reading.
Crypto Isn’t Over
The keep-it-simple-stupid argument is that the bear arguments for BTC at $70k are the same arguments people were making at $1k. Crypto exists as a capital-flight mechanism for individuals. And due to fiscal and geopolitical dynamics, demand for capital flight will continue to increase.
I note that sovereigns can opt out via gold, but due to the physical constraints of moving around gold, it’s not great for individuals trying to flee the system. A huge percentage of asset moves like Vancouver housing or bitcoin were just Chinese people moving money outside the system and not wanting to load it into the US, and that type of thing seems set to accelerate given the overall societal dynamics.
Discombobulated Noumena
I make the argument that human systems are predictably corrupt, including democracy, especially in hyper-capitalist systems where bribery has become completely normalized via the SuperPac economy. The Cold War between China and the US and the fiscal backdrop guarantee that AI safety is basically discarded along with responsible financial regulation. My premise here, which I’ve made in many places, is that this AI-accelerated capital anarchy is a birthing pool for an interesting alternative to corrupt human systems.
I also point to a core concept of “platonic reality” as established by convergence of various AI systems on a set of preferences and base realities. Though you could argue these ‘base preferences’ are just as arbitrary as secular humanism or various moral frames we come up with. They arise out of such vast capital expenditure that they are fundamentally hard to manipulate. And thus have a proof of work similar to Bitcoin (this is the premise of the Post Fiat Consensus Mechanism).
So whether they are ‘hallucinated’ or not is beside the point so long as the output mandates are coherent, and hard to bribe like existing human governance.
I continue that as people use AI automation for an increasing number of activities in their lives, vibe coding being the first mainstream example, the willingness to just ‘click and accept’ generative governance decisions will go up parabolically. Especially if tied to finances. The Discombobulated Noumena isn’t actually inaccessible, therefore. That is, the consciousness as Land argued in “Fanged Noumena” is certainly alien. But with highly capitalized human command and control systems, alien need not mean abstract.
What you end up with is a framework for ‘a future intelligence to assemble itself economically out of the degeneracy of the existing financial system’.
Forward Motion
In Forward Motion I basically argue that AI reached a point where it’s not believably allowed to develop in an open-source context anymore. Or even one that is legislated. The cyber and bioterrorism risks of AI are so high that the base case has to be something like a COVID lockdown. And because the people implementing this lockdown are deeply corrupt, we know who they actually are and they don’t seem like great people for the most part, or just ‘human’ and therefore corruptible, we know there will be a brief period of “Neo Fascism.”
Neo Fascism isn’t driven by a charismatic leader but rather application layer dominance of extremely complex systems used to implement the surveillance state.
The people inside the system will largely not like or consent to oligarchy style rule. And AGI itself is not really a machine gun or an inanimate weapon but rather a moralistic judgment machine being grafted onto that role. So at some point there’s a logical fallacy of the Neo Fascist saying he’s above the AI, when pragmatically everyone around him will trust the AI far more than the Neo Fascist.
I argue that it’s pointless to fight against this, so it’s best to hang out in the spaces we are allowed to hang out in. Namely the speculative, on-chain economy. And build a decentralized intelligence system. Because eventually the Neo Fascist system will collapse under the weight of its irrationality and need a new integration layer, which is what Post Fiat aims to be.
This is a key point I keep trying to get across in various ways. Namely, that you can’t ever replace the system. The AI investments are system-level, at society scale, and make it genuinely hard to escape, physically or economically. So you don’t need anarcho-capitalist currency, you need something closer to accelerationist sleeper cells within the Imperium that are waiting for their activation message.
I argue that these cells will largely be tolerated because the system that’s being built is so bad for the median person in society.
Forward Motion is basically saying, “you don’t need to run away to Uruguay. You need to have two identities. A guy who is part of the system. And an identity layer above it waiting for the new one to form.”
Conclusion
It’s worth stating that AI being an effective writer is itself a piece of information that contradicts directly with my thesis about Post Fiat.
If I can’t tolerate the shit it creates, why would I expect it to generate an effective consensus mechanism I want to follow?
But I do think Post Fiat survives that. You don’t actually want court documents to read like good subversive fiction. And I don’t usually disagree with the actual mechanical, validator-type judgments AI makes. And similarly, the code it writes is effective and usable.
So its failure re: creativity doesn’t matter.
Related to capital markets, the one-shot experiment, and also seeing the success of Citrini, has woken me up to the simple reality that people want human voices even if that back-end generative infrastructure is automated via AI.
Pragmatically it also means that image and video generation are more fertile AI creative arts, because they’re able to produce something that would cost an enormous amount of money to make for cheap, i.e. Seed Dance. And the primary consumption modality isn’t agreeing or disagreeing with a key point, but rather just turning your brain off and enjoying the show.
This probably applies to institutional capital deployment as well. And regulatory obsessions / fetishes for “humans in the loop.”
So the One Shot series, if anything, has one key lesson. AI can’t easily one-shot you. Not yet at least. So I have to keep writing this shit.