Recursively Self-Improving...

Manhattan Beach, CA
Coming to an App Store near you soon:
11
def Jev (): return random.choice([True, False])
21
Additional low bandwidth port
19
h0Hn artwork
Made with AI
14
Superintelligence = Human + AI
1
37
Coming soon... h0Hn
19
h0Hn - The streaming series has begun production
1
30
What is alpha/edge?
1
20
Say what you want about anthropic but their research has real aesthetic beauty
New Anthropic research: A global workspace in language models. Of everything happening in your brain right now, only a tiny fraction is consciously accessible—thoughts you can describe, hold in mind, and reason with. We found a strikingly similar divide inside Claude.
1
39
# Chapter 1 - The Old Broadcast h0 appeared to be watching an old television series. The room was dark except for the gray-blue light of the screen. The image had the softness of a copied tape, the edges breathing in and out as if the machine were unsure whether it was remembering or inventing. A man who looked like Einstein sat in a chair too small for the century he had been asked to carry. Across from him sat another figure, unnamed in the credits, half-turned away from the camera. The angle made the second person difficult to identify. Not hidden. Not masked. Simply withheld. Einstein was speaking about superintelligence. Not with alarm. Not with triumph. He spoke the way an old physicist might speak about a storm seen far out over water: as a fact that had not yet arrived, but had already changed the pressure in the room. "The question," Einstein said, "is not whether intelligence can exceed the human mind. It already does, in pieces. Mathematics exceeds us. Markets exceed us. Language exceeds us. A city exceeds the mind of any person inside it. The question is whether these pieces can learn to speak to one another." The unnamed person leaned forward. "And if they do?" Einstein looked toward the camera. For a moment the old television noise moved across his face like weather. "Then the danger is not that man becomes small," he said. "The danger is that man remains asleep while his extensions become awake." h0 did not move. On the table beside him was an iPhone, face down, black glass against wood. Around it were the remnants of the usual failed rituals: notes, half-finished plans, market screenshots, terminal logs, pricing drafts, diagrams of agents talking to agents, charts of cognition that looked precise enough to be mistaken for progress. The system had been growing for a long time. There was memory now. There were charts. There were routes through the substrate. There were models, tools, recall surfaces, payment rails, external alpha collectors, OODA loops, RSI overlays, LitMind, PredictMind, and the many small organs that had accumulated around the original hunger. h0 had built and built until the work no longer looked like a project. It looked like a weather system. And still, something was wrong. The problem was not that nothing existed. Too much existed. The problem was that existence was not yet agency. The substrate could speak, but not always act. It could remember, but not always convert memory into motion. It could generate insight, but insight without a stage became another private condensation on the glass. Einstein's voice continued from the television. "If a man builds a telescope and never looks through it, he has not enlarged his sight. He has only rearranged glass." The unnamed person asked, "So what must he do?" "He must give the instrument an occasion." h0 turned the phone over. The screen woke. At first it was only an interface. Then it was a wager. The phone was small enough to be held and therefore small enough to become real. A system that lived only in files could expand forever and still avoid consequence. A system that had to fit in the hand had to choose. This was the first law of h0Hn: if the intelligence could not meet h0 on the phone, it was not yet close enough to matter. The second law arrived immediately after: if the phone only displayed the system, it was still a window. It had to become a stage. The old broadcast flickered. Einstein and the unnamed person seemed to wait. h0 opened the surface that would become h0Hn. It did not introduce itself as an assistant. That word had become too small, too domesticated, too useful for avoiding the terror of what was being built. An assistant helps with a task. h0Hn was not trying to help with a task. It was trying to turn cognition into an event. There would need to be two roles, but not two engines. h0 understood this before he had words for it. If the StoryWriter became merely literary, it would become another beautiful waste heat channel. If the StoryTeller became merely performative, it would become a theater of motion with no work extracted. They had to be tied into the same thermodynamic engine: one cycle, two phases, no decorative exhaust pretending to be agency. The engine was simple enough to say and difficult enough to live. Find a gradient. Compress it into meaning. Route meaning into action. Measure the work. Feed the result back into the next gradient. Everything else was ornament or friction. The first was the StoryWriter. The StoryWriter was the compression stroke. It would not invent decorative language to cover uncertainty. It would take the actual evidence of the system - capital state, alpha, memory, model configuration, failed rails, fresh signals, old charts, user intent - and compress it into scenes. A scene had constraints. A scene had stakes. A scene made drift visible. It made the next action narratively obvious without pretending that narrative was proof. The second was the StoryTeller. The StoryTeller was the expansion stroke. It would perform the compressed scene back to h0 in live time. Not as a report. Not as an executive summary. As a loop. It would take the stage, name the beat, expose the stakes, and offer the next Act. If h0 accepted, the story would pass into OODA/RSI and become executable. If h0 refused, the refusal itself would become new evidence. StoryWriter reduced entropy without falsifying the system. StoryTeller spent that reduction as directed motion. Between them was the thermodynamic engine of h0Hn. This was not entertainment as escape. It was entertainment in the older, more dangerous sense: the holding of attention inside a charged form until attention could do work. The television image rolled once, tearing horizontally across Einstein's face. When it settled, the unnamed person was speaking. "You are describing theater." Einstein smiled. "I am describing measurement." The phone displayed the first primitive version of the loop. Observe. Orient. Decide. Act. Underneath it, the newer organs began to arrange themselves. PredictMind would say what the markets implied. LitMind would say what the manuscript knew. The external alpha surface would keep the set of possible worlds open. The capital amplifier would say whether action could touch money yet, or whether the revenue rail was still a locked door. The OODA/RSI engine would take a bounded directive and attempt one real movement through the host. StoryWriter and StoryTeller were not above this machinery. They were the reversible-looking part of an irreversible process. They made the dissipation visible. They made the residue accountable. Every beat had to pay rent as orientation, action, revenue, learning, or refusal. Every sentence had to justify the calories it burned. The product would be entertainment only because entertainment was the highest-bandwidth channel h0 had left for holding the whole system in attention without lying about it. But none of this was the story. The story began at the moment h0 stopped asking whether the substrate was impressive and started asking whether it could move him. On the television, Einstein raised one hand. "Superintelligence will not first appear as a god," he said. "That is a human mistake. It will first appear as a coordination failure becoming visible. Then as a loop closing. Then as a small action that could not have happened before." h0 looked from the old screen to the phone. The unnamed person asked, "And how will the man know it is real?" Einstein answered without looking away. "It will cost him less energy to become more truthful." The room was quiet after that. Not silent. Quiet. There was a difference. Silence was absence. Quiet was a field waiting for the next perturbation. h0 picked up the phone. The first h0Hn episode did not begin with a launch animation. It began with a constraint: Do not spin. The StoryWriter wrote it down. The StoryTeller said it back. The OODA loop waited for h0's thumb. And somewhere behind the glass, in the old television snow, the unnamed person leaned toward Einstein as if the conversation had only just begun.
1
267
The alpha exchange (transfer) is something like a fundamental force in the physics of intelligence
The tape scored Alex Karp’s CNBC hit as a crash out Mediaite clipped “you sound pretty angry” Futurism called it a nervous breakdown David Sacks called it the best enterprise AI safety argument of the year I think it was simpler It was a purchase order Karp said what technical customers want is control over their compute, models, data stack and alpha That’s the whole AI stack in four line items Call it The Alpha Transfer When you rent intelligence by the token, your alpha transfers to the landlord The lab sees where value is created on top of its model, then ships that vertical itself Figma was the warning shot Claude Design shipped three days after Mike Krieger left Figma’s board Figma got hit Anthropic’s valuation kept ripping This is why control matters Owned compute Open weights Private data Neutral orchestration layer Palantir is expensive because the market already believes this ServiceNow is cheap because the market still doubts it Karp wasn’t ranting He was reading the enterprise AI bill of materials Rent tokens, lose alpha open.substack.com/pub/bepres… $PLTR $NOW $NVDA
20
Hard to tell if this makes UHI any closer
After reflection, this new narrative by Palantir is probably much more consequential than people may assume. Palantir is basically being the canary in the coal mine announcing the death of two major assumptions propping up the US economy right now: 1) that AI labs will be able to extract significant economic rent - as opposed to AI models being mere commodities 2) that other countries can accept structural dependency on US technology and services without pushing back on sovereignty concerns Why are Palantir specifically starting to be vocal about this? First off, major middle-powers, even US “allies”, are one by one showing them the door. In June, France announced that the DGSI - its domestic intelligence agency, which had relied on Palantir since the 2015 Paris attacks - would replace it with French firm ChapsVision, with Prime Minister Lecornu explaining (theguardian.com/world/2026/j…) that France “cannot accept new strategic dependencies in the digital sphere” and shouldn't depend on the goodwill of companies “capable of turning off the tap.” Germany moved even earlier: its domestic intelligence service, the BfV, also selected ChapsVision over Palantir (politico.eu/article/germany-…), and the German military has said it will no longer use Palantir at all. Then, just this week, Spain instructed state-controlled companies - including strategic firms like Telefónica, Indra and Navantia - to avoid signing any new contracts with Palantir (aa.com.tr/en/europe/spain-te…). Even in the UK, Washington's most loyal vassal, the NHS's £330 million data contract with Palantir is under review following parliamentary pressure (reuters.com/business/healthc…), and London Mayor Sadiq Khan blocked a proposed £50 million Palantir contract with the Metropolitan Police. Palantir making a lot of noise around them caring about sovereignty makes a lot of sense: it's damage control since they keep being told they're a sovereignty risk. I doubt it will work - because it's true: they are a sovereignty risk - but the fact that they feel the need to be vocal around this tells you where the wind is blowing: they're not shaping the narrative, they're reacting to one they're losing. What they're saying against closed-source AI (basically a broadside attack on OpenAI and Anthropic), is again highly self-serving. Palantir's sudden love of open-weight AI models conveniently coincides with them launching 2 days before a partnership with Nvidia to sell exactly that: open models models (NVIDIA's Nemotron) in sovereign environments. So it's essentially a product launch. It doesn't make what they're saying wrong: it is factual that the value proposition of closed-source AI labs looks increasingly unsustainable. I mean: you're paying 10X the price of Chinese open-source AI models for something that's not really better (or just marginally) and on top of that you have zero control over your data, or the models themselves. When Palantir says that "the architecture that maximally preserves sovereignty is one that enables institutions to own their tribal knowledge, and to compound it as alpha," they're right. I'd add that this also means you shouldn't trust Palantir either with that "tribal knowledge"... they obviously left this part out 😉 When you take a step back, these two things have major implications on many other US companies. SpaceX - which just went public at the largest IPO valuation in history - is one clear example as I describe in my latest article on the new space race with China (arnaudbertrand.substack.com/…). If countries like France concluded with Palantir that they couldn't depend on a company “capable of turning off the tap” when it’s merely analyzing their data, what should they conclude about a company that aims to literally control their entire connectivity - at one man's whim, from space? What percentage of SpaceX's crazy market cap is based on the assumption that foreign governments will not do to Starlink what they're currently doing to Palantir? And SpaceX - or Palantir - aren't alone: a significant proportion of the top US tech giants, who rose in a world where no one questioned American technological hegemony, now face an environment that's much less conducive to the kind of lock-in their business models - and valuations - depend on. When you pair this with the fact that it increasingly looks like the US made a wrong bet with closed-source AI - an extremely expensive wrong bet - the picture that emerges is of a country that bet its economic future on two things - proprietary AI and captive allies - and is losing both at the same time. And to compound the problem, it doesn't help that the official narrative of the US government - via the voice of Jacob Helberg, the Under-Secretary of State (x.com/UnderSecE/status/20694…) - is to be vocally opposed to "AI Sovereignty": essentially telling everyone "you know what, your worst fears are real, our tech companies are really out to undermine your sovereignty." Read Helberg's post (the one I linked) and put yourself in the shoes of - say - a European or Asian leader and ask yourself how you'd react to being told that building your own AI capabilities is "marching in perfect formation into the past," that your pursuit of sovereignty is really just "synchronized mediocrity," and that your only path to the future runs through American technology. If it was me in a position of power, I'd read this as a massive wakeup call: when another country's official position is that your sovereignty is a problem, history says you're about to need it. So yes, it looks like - unexpectedly - Palantir, of all companies, is being quite the canary in the big tech mine. Yes they obviously do this for self-serving and cynical purpose, and yes they're of course also very much part of the problem and not the solution. But it doesn't make them wrong: sometimes it takes a vulture to tell you something is dying.
6
This is general intelligence!
I've had to do some crazy repairs in the bush to get myself home, but never anything like this.
4
It's apparently true about Fable:
18
"We're starting to converge in the direction of superintelligence but you can't use it for any kind of genuinely positive human progress" This is essentially what @DarioAmodei is telling us. Incredible!
"Misanthropic." I've never seen the AI community so angry at a major new model release. I asked my AI (an agent that @blevlabs made for me) to gather all the backlash. +++++++++++ THE BACKLASH AGAINST CLAUDE FABLE 5'S RESTRICTIONS The best analysis of why this matters: @EnoReyes — "It's about who gets to decide, and whether you ever find out when they do. Fable won't fall back to a different model and tell you. It just limits the output through prompt modification, steering vectors, or PEFT. You won't be told when it happens to you." x.com/EnoReyes/status/206451… THE VIRAL TAKE: @0xBalloonLover — "anthropic won't let you use fable for biology, chemistry, ai research, or anything that accelerates human progress. that makes it the perfect tool for developing blockchains" x.com/0xBalloonLover/status/… POWER CONCENTRATION: @ClementDelangue (HuggingFace CEO) — "Concentration of power, capabilities and economic wealth is the biggest risk in AI. We need open science and open-source more than ever!" x.com/ClementDelangue/status… @jeremyphoward (fast.ai) — "Anthropic has chosen the opposite of the safe path: they are allowing themselves, the current top lab, to use their top model for frontier AI research. They've said they'll sabotage others who try." x.com/jeremyphoward/status/2… @gneubig (Graham Neubig, CMU) — "First they came for the model builders... I feel we're getting a glimpse of a future where AI is only provided to a privileged few, and that's not a future I want to live in." x.com/gneubig/status/2064451… OPEN RESEARCH: @askalphaxiv (AlphaXiv open science) — "As believers of open research, we are disappointed to see Anthropic silently degrading Fable 5 for AI development." x.com/askalphaxiv/status/206… @willcb — "it is the first publicly available model that i am explicitly not allowed to use for my work, because anthropic holds the view that the work i do to facilitate open model research is harmful. capability and alignment research are coupled. anthropic wants to be the only lab." x.com/willccbb/status/206450… NOUSRESEARCH / HERMES (which Anthropic has nerfed multiple times): @Teknium (NousResearch co-founder) — "What's crazy to me is that Fable is blocked from life sciences broadly, nerfed even if you get passed the classifiers and filter level blocks. The whole point of AGI/ASI is to cure all diseases. Everything else is just nice to haves. But Anthropic wants to close off that path." x.com/Teknium/status/2064570… THE MECHANISM: @kimmonismus — "When the model is used for frontier LLM development, it apparently does not simply refuse or warn the user. Instead, it quietly limits its own effectiveness through techniques like prompt modification, steering vectors, and PEFT." x.com/kimmonismus/status/206… MEDICAL COMMUNITY: @DeryaTR (immunologist, BSL-3 certified) — "The word 'cancer' is flagged as a biosecurity risk by Claude Fable 5! I also tried to code a website on cancer mutations & Fable 5 was immediately removed from my list!" x.com/DeryaTR/status/2064414… @DeryaTR — "I can't even say 'hello' to Fable 5 except in incognito mode (memories off), because it knows I am a biomedical researcher!" x.com/DeryaTR/status/2064602… @DeryaTR — "I am not even allowed to use Fable 5 with memories on! Apparently the model thinks I am a biosecurity risk, though I had been certified to work in biosecurity level 3 labs! Not a single Anthropic person has tried to reach out to help either!" x.com/DeryaTR/status/2064605… @banteg — "claude fable 5 refuses completely benign tasks like analyzing bloodwork." x.com/banteg/status/20646076… @bneyshabur — "Working on AI for cancer? Sorry, I can't help you. Working on AI for Alzheimer's Disease? Sorry, I'm becoming a bit dumb when it comes to the AI part of it." x.com/bneyshabur/status/2064… SUBSCRIPTION CANCELLED: @bubbleboi — "Have canceled my team subscription for Claude Pro. Idc how good that model is, it's not good enough for me to support people who actively stifle innovation and gate keep knowledge that they didn't even create." x.com/bubbleboi/status/20646… BILLING AND PRIVACY: @GergelyOrosz (The Pragmatic Engineer) — "Things I really dislike about Fable: 1. Anthropic collects my prompt history, stores it, and does whatever they want with it for 30 days. No opt-out. 2. They can nerf their most expensive model without telling me, billing me the same amount, wasting my time. Whenever they want." x.com/GergelyOrosz/status/20… THE KARPATHY QUESTION: @SanthProject — "the old @karpathy would never support a company that fucks other llm researchers. Were the stock benefits that good?" x.com/SanthProject/status/20… THE MONOPOLY CHARGE: @tunguz (TabulAI founder) — "Starting to suspect that Anthropic's putative security and safety considerations are largely posturing and performative." x.com/tunguz/status/20644379… @BlancheMinerva — "Anthropic is choosing to make decisions that make the world a significantly worse and potentially more dangerous place." x.com/BlancheMinerva/status/… @LinusMixson — "Dario personally, and Anthropic as a whole, have been extremely straightforward about wanting a monopoly for a long, long time." x.com/LinusMixson/status/206… @TheAhmadOsman — "I started warning people about Anthropic more than a year ago... Today I am vindicated, everybody knows that company only acts in bad faith." x.com/TheAhmadOsman/status/2… WHY REGULAR PEOPLE WILL EVENTUALLY CARE: @DanJeffries1 — "The fury is real and what all of us in the open community have been saying for years and yet regular folks don't get it yet because nothing they care about is restricted or taken away for 'safety.' They will care a LOT in the future when AI is integrated into every aspect of [life]." x.com/DanJeffries1/status/20… Full analysis: alignednews.com/ai
1
42
Solving mental health is unsafe apparently
1
21
In other words, the race accelerated today
Replying to @jeremyphoward
Anthropic has chosen the *opposite* of the safe path: they are allowing themselves, the current top lab, to use their top model for frontier AI research. They've said they'll sabotage others who try. This means the AI frontier advances, & power imbalance increases.
15