The Superintelligence Rename: Narrative Inflation and the Verifiability Gap

Last Tuesday I did what I always do when a story smells wrong: I went to the primary source. A post was moving through my feed claiming that a sitting U.S. president had signed an executive order forcing federal agencies to stop calling the field "artificial intelligence" and start calling it "SI" — Superintelligence. The same thread added a second beat: Tesla, in apparent lockstep, had renamed its AI account to "TeslaSI." Within forty-eight hours the claim had reached three group chats and one newsletter that genuinely should know better.

So I opened the Federal Register. Nothing. I opened the White House site. Nothing. I checked the X account in question, which is the closest thing we have to a live audit trail for corporate naming. Unverifiable. Seven days of digging produced a story with no floor under it — and a far more interesting question than the one everyone was asking. Why did we want it to be true? That question is the whole article.

The story landed on top of a real and genuinely consequential fight over AI vocabulary. For three years the most powerful labs on earth have been competing not only on models but on names. OpenAI pushed "AGI" — artificial general intelligence — into the mainstream and built an entire governance philosophy around the phrase. Ilya Sutskever left to found Safe Superintelligence, Inc., and hung the word "Superintelligence" on the door. xAI's stated mission is to "understand the true nature of the universe," a phrase so grand it functions as a naming strategy in itself. Google DeepMind traded in "frontier AI" and, increasingly, "ASI" — artificial superintelligence. Each of these labels is a claim about the future, and each is also a bid for the right to define the terms of the debate.

Now understand what a naming war actually is. It is not marketing fluff layered on top of technology. In a capital-intensive, attention-scarce industry, the vocabulary is the product roadmap for the public imagination. Whoever owns the word owns the frame; whoever owns the frame shapes which questions get asked, which risks get funded, and which regulators show up. When OpenAI says "AGI," it is staking a claim about scale and safety. When SSI says "Superintelligence," it is staking a claim about destiny. When a viral post says a president renamed AI to SI, it is staking a claim about inevitability — that the thing has already arrived, that the grown-ups have already agreed on what to call it, and that you should adjust your worldview accordingly.

I have spent twenty-seven years watching this movie, and I have seen it before. Not in AI — in crypto. The blockchain industry invented narrative inflation as an art form, and we are now watching the AI industry rediscover the technique in real time, with better funding and worse epistemics. In 2017, I coded the initial smart contracts for CapeHorizon, a decentralized community governance protocol meant to fund local creative arts in Cape Town. I raised $120,000 in ETH from five hundred early adopters who showed up to meetups in Woodstock because they believed in the story. The story was beautiful. The gas fee management was not. When Ethereum congestion hit in November 2017, the whole thing seized up, and I learned the first lesson of my career: a narrative is not infrastructure. Decentralization is a promise the machine has to keep, not a slogan the founder gets to repeat.

That lesson is why I stopped trusting headlines and started trusting ledgers. And it is why the "SI" story interests me — not as news, because it is almost certainly not news, but as a diagnostic. The story is a stress test for how AI, and by extension crypto, handles its own mythology. And the patient is running a fever.

The first thing to understand is that "AI" and "SI" are not two words for the same thing. They are not even two words for two things at the same level. This is a category error dressed up as a policy change, and the error is instructive.

The industry has, for years, used a rough ladder to describe capability. At the bottom is ANI — artificial narrow intelligence — the systems we actually have: models that are spectacular within a domain and helpless outside it. In the middle is AGI — artificial general intelligence — a hypothetical system that can match or exceed humans across essentially all cognitive tasks. At the top is ASI or SI — artificial superintelligence — a hypothetical system that exceeds the best human minds in every relevant dimension, by a margin we cannot currently compute.

Notice the structure. ANI is a description of a class of technology. AGI and ASI are descriptions of capability levels that no system has reached. "AI" is the umbrella term for the entire field — the container. "SI" is a rung on the ladder inside that container, and it is a rung nobody has stood on. To replace "AI" with "SI" is like replacing the word "vehicles" with the word "supersonic." It is not a renaming. It is a category collapse. You have taken the noun for the domain and swapped in the adjective for an aspiration, and in doing so you have smuggled a claim — that we have arrived — into a label that is supposed to be neutral.

This is why the story, if taken at face value, would be alarming even though it sounds like a joke. An executive order that redefines a technical field as a capability threshold is not a bureaucratic formality. It is a government, in effect, asserting that the threshold has been crossed. Every downstream consequence follows from that smuggled claim: funding priorities, safety frameworks, public expectation, the entire emotional register in which citizens think about the technology. If you rename the field after the summit, you have told everyone the climb is over.

Here is the part that makes me, as a Web3 person, sit up straight. Crypto did exactly this, and we are still paying for it. We renamed things that were aspirational as though they were descriptive. We called tokens "governance tokens" when the governance was a snapshot vote that three wallets controlled. We called chains "decentralized" when the sequencer was a single company's server and the upgrade keys sat in a multisig. We called lending protocols "risk-free yield" when the risk was simply hidden in a composability chain that nobody had stress-tested. The words ran years ahead of the machines, and when the words collapsed, they took real money with them.

I lived that collapse in 2020. During DeFi Summer, my ENFP brain lit up the way it always does in the presence of possibility. I joined three yield-farming protocols at once, chasing APYs above one hundred percent, and I put $50,000 of personal savings into a mix of Uniswap LP positions and novel lending platforms. I made about $15,000. I also discovered, by accident, the composability risk of leveraged yield strategies — the way one protocol's failure cascades through another's collateral, then another's oracle, until the whole structure is revealed as a single bet wearing many costumes. The lesson was not that the yields were fake. The lesson was that the yields were real and the labels were fake. The number on the screen was accurate. The word "safe" was not.

That is narrative inflation: a gap between the word and the machine, and a market that prices the word. AI is now running the same experiment at a much larger scale. When a lab's valuation rests on the word "Superintelligence" rather than on a shipped capability, the word is doing the work of the product. SSI reportedly raised at a valuation north of thirty billion dollars on the strength of a thesis and a team — a thesis that is, by the founders' own framing, about a system that does not yet exist. That is not fraud. It is a bet on a narrative, and bets on narratives are legitimate. But it is worth naming what is happening: the label has become the asset.

Now put the "SI rename" story back into that frame. The story is fake — or, more precisely, its verifiability is absent, which for a news consumer should be the same thing. But its plausibility is real. It is plausible precisely because the industry has spent years inflating its vocabulary to the point where a presidential renaming of a field after its unreached summit does not sound absurd. The satire works only because the straight version is already everywhere. The joke is not that the story is outrageous. The joke is that it is almost believable. And anything almost believable is, functionally, a lie waiting to be repeated.

I have a technical objection to the whole genre, and it comes from the thing I actually build. Verifiability. In blockchain, we have one genuinely novel contribution to the world of information, and it is not decentralization — it is the ability to check a claim without trusting the claimant. A block explorer does not ask you to believe a founder. It shows you the transaction. The whole architecture is a machine for converting trust into inspection. That is what "code is law" was always supposed to mean, before the slogan got captured: not that machines should rule us, but that claims should be checkable by anyone.

And yet the crypto industry, which built the greatest verification machine in history, spent a decade inventing labels it could not verify. We called things "Layer 2" that were, in practice, custodial bridges with a marketing department. We called things "Bitcoin Layer 2" that were Ethereum projects wearing a Bitcoin costume, hoping the word would do the work the architecture could not. I have audited enough of these to say it plainly: when a team rebrands an Ethereum rollup as a "Bitcoin Layer 2," what has changed is the pitch deck, not the security model. The real Bitcoin community does not acknowledge these projects, and it is right not to. You cannot rename your way into a security guarantee.

This is the bridge between the two stories, and it is the core insight of this piece: AI and crypto are suffering from the same disease, which is that narrative has decoupled from verifiability, and both industries are now paying a premium for words instead of machines. In AI, the word is "Superintelligence," and it is being used to sell valuations. In crypto, the word is "decentralized," and it was used to sell tokens. In both cases, the fix is not better words. The fix is a way to check.

So let me get concrete about what checking looks like, because this is where I think the two industries actually need each other, and where the real opportunity lives.

When I pivoted during the 2022 bear market, I did not pivot toward price. My portfolio was down seventy percent and my mood was down more. I pivoted toward cryptographic truth. I spent six months studying zero-knowledge proofs, largely through the work coming out of Succinct Labs, and I published a three-part explainer series called "Privacy in a Transparent World" that eventually reached about fifty thousand readers. That research changed how I think about verification. ZK proofs are not just a privacy tool. They are a proof-of-honesty tool. They let you demonstrate that a statement is true without revealing why it is true. You can prove you ran the computation correctly without exposing the computation. You can prove a model produced a given output from a given input without handing over the weights.

That is the primitive AI desperately needs and does not yet have at scale. Right now, when a lab tells you its model is safe, or that a benchmark score is real, or that a generated image is authentic, you are trusting a press release. When a news account tells you a president renamed a field, you are trusting a screenshot. The entire information economy of AI runs on assertions that no one can independently check, and the result is exactly what you would predict: a landscape in which satire, marketing, and genuine news are visually indistinguishable, and in which the burden of skepticism falls entirely on the reader.

I tried to build the fix. In 2026 I launched TruthChain, a community-driven project to authenticate AI-generated content using on-chain proofs, backed by $200,000 in community funds and a team of AI researchers and cryptographers. My role was not to write the proving system — I am a finance guy who learned Solidity the hard way — but to articulate the ethical framework, to translate between the people who build the cryptography and the people who write the rules. We onboarded ten thousand users looking for verified sources. It was the most meaningful work I have done, and it taught me something uncomfortable: verification is a product, and like every product, it can be faked.

Which brings me to the part of this argument that cuts against my own instincts, the part I have to say because the alternative is to become the thing I am criticizing.

Here is the blind spot in the "just verify everything" gospel. Verification is not free, and it is not automatically honest. It has three failure modes, and the AI and crypto industries are already tripping all three.

First, you can verify the wrong thing. An on-chain proof that a piece of content was signed by a given key tells you nothing about whether the content is true. It proves provenance, not accuracy. I can mint an NFT that attests I said the sky is green, and the chain will faithfully record that I said it. The ledger is a witness, not a judge. Crypto learned this the hard way with "verified" contracts that were verified to be malicious. AI is about to learn it with "authenticated" content that is authenticated misinformation. A cryptographic signature on a lie is still a lie, and it now carries a patina of trust it did not earn.

Second, verification can become theater. This is the subtler trap, and it is the one that worries me most because I have seen it up close. "Build in public" is a beautiful principle that is trivially gamed. A team can publish a dashboard, stream a dev call, post an audit — and still be running a narrative that the artifacts are designed to support rather than to test. The ritual of transparency substitutes for the substance of it. In crypto we call this a "transparency theater" problem, and it is why so many audited protocols still blew up. The audit checked the code that was written, not the intent behind it. In AI, the equivalent is the safety report that is published to satisfy a regulator while the actual training run proceeds on a different set of assumptions. The word "verified" gets stamped on the box, and the box is still a box.

Third, and this is the pragmatism test I keep applying to every claim I read: does verifiability actually change behavior, or does it just add a cost layer that honest actors pay and dishonest actors route around? If the only people who bear the cost of verification are the ones already telling the truth, then verification is a tax on honesty, not a defense against lies. I have watched this happen in KYC, in audit culture, in compliance regimes across finance. The good actors fill out the forms. The bad actors find the gap. And the gap is always there, because the map of what can be verified is never the same as the territory of what matters.

So when I say the fix for narrative inflation is verifiability, I have to be honest about the limits of my own prescription. Verifiability raises the cost of lying. It does not eliminate lying. It shifts the battleground from "can you prove it?" to "did you prove the right thing, and does anyone care?" The last part is the killer. A verification system only works if people actually look. And the entire economics of the attention economy are built on the assumption that they won't.

That is the deepest reason the "SI" story spread. Not because it was checkable — it was, trivially, and it failed every check — but because checking is work, and sharing is fun. The Federal Register is one tab away from anyone who wants it. Almost nobody opened it. The narrative was cheaper than the truth, and the attention market rewarded the cheaper good. This is not an AI problem or a crypto problem. It is the substrate problem on which both industries float.

And yet I am not a pessimist, and this is where my ENFP wiring kicks in against my own analysis. Embrace the volatility, find the signal. The volatility here is the flood of fake, inflated, and satirical content that will define the next decade of the information economy. The signal is that this flood creates a genuine, permanent, and underserved market: not for more content, but for trusted verification of content. Not for louder words, but for checkable machines. That market does not exist yet at scale, which means the people who build it are not competing with incumbents. They are competing with inertia, which is a much weaker opponent.

I want to be specific about where I think the signal is, because vagueness is its own form of narrative inflation.

One. Provenance infrastructure for generated media. The thing TruthChain tried to be, and that the industry still needs: a standard way to attach a verifiable origin to a piece of content, so that a reader can distinguish a real photograph from a generated one without trusting the platform that shows it to them. The technical primitive exists — signatures, hashes, ZK proofs. The missing piece is adoption, and adoption requires a distribution partner with a reason to care. Regulators who are about to be flooded with synthetic evidence are that partner.

Two. On-chain attestation for AI claims. When a lab claims a benchmark score, a safety property, or a training-data provenance, that claim should be verifiable by a third party without the lab revealing its weights. ZK proofs make this possible in principle. Nobody has built the standard. This is a real gap, and it is exactly the kind of problem that sits at the intersection of the two industries I live in.

Three. The deflation trade. Every inflated narrative eventually deflates, and the people who profit are the ones who can tell the difference between a word and a machine before the market can. This is not a call to short anything. It is a call to build an analytical practice: when you read a claim about AI or crypto, ask three questions. What is the primary source? What does the claim actually assert, stripped of adjectives? And does the artifact — the code, the ledger, the benchmark, the paper — support the word, or merely coexist with it? I have made a career out of those three questions, and they have saved me more money than any model.

Let me be clear about where the "SI rename" story fits in all this, because I do not want to leave the impression that I think it is harmless. I think it is a small symptom of a large condition, and small symptoms of large conditions deserve attention precisely because they are cheap to ignore. The story is almost certainly satirical or fabricated. Its source, a name that reads like a pun on "breaking AI news," has no editorial accountability. It cites no primary document. It describes a policy that does not exist and a corporate action that does not follow from that policy. By every standard of evidence I was trained to apply in finance, it fails. And it still traveled, because it told a story people were already primed to believe: that the superintelligence era is here, that the powerful have already agreed on it, and that the rest of us are just catching up.

That primed belief is the actual asset being traded. Not the story — the belief. And the belief has a price, expressed in valuations, in funding rounds, in the willingness of serious people to repeat unserious claims. The AI industry is monetizing a gap between its vocabulary and its machines. The crypto industry did the same thing, and the bill came due in 2018 and again in 2022. I do not know when the AI bill comes due. I know it is being written.

I also know, from twenty-seven years of watching cycles, that the bill is never paid by the people who wrote it. It is paid by the retail reader who believed the word, the small investor who priced the narrative, the community that moved in before the infrastructure was ready. I built CapeHorizon and watched five hundred believers lose their faith in decentralization because I could not manage gas fees. I launched a generative art collection that sold two hundred pieces in forty-eight hours for $80,000 and then stagnated because I could not sustain the value proposition past the hype. In both cases, the word outran the machine, and the community paid for my vocabulary.

That is why I write this way now — short sentences for the hard facts, long ones for the human stakes. The hard fact is that the "SI" story is unverifiable and should be discarded. The human stake is that the story is unverifiable and should be discarded, and yet it will not be, because the industry has trained its audience to prefer the word to the machine. The hard fact is that no executive order renamed anything. The human stake is that a great many intelligent people will spend the next year arguing about a renaming that never happened, while the actual renaming — the slow, quiet, market-driven inflation of AI vocabulary — proceeds without a single headline.

So what do we do with this? Not panic, and not cynicism. Both are forms of surrender. The productive response is the one I keep coming back to, the one that cost me the most to learn: build in public, live in truth. Publish the primary source. Show the ledger. Open the code. And when someone hands you a story with a president and a billionaire and a spooky acronym, open one more tab before you open your mouth. That single habit, repeated at scale, is worth more than any regulatory framework anyone could write.

Here is the future-back view I want to leave you with. Imagine the information ecosystem five years from now, when synthetic content is the default and the cost of generating a convincing claim has fallen to zero. In that world, the scarce resource is not intelligence — it will be abundant, cheap, and everywhere. The scarce resource is verifiable truth: the ability to know, without trusting anyone, that a thing is what it says it is. That is the world blockchain was built for, and it is the world AI is dragging us into whether we like it or not. The two industries that spent the last decade inflating their vocabularies will spend the next one being asked to cash the words in. Some will have the machines to back them up. Some will not. And the difference between them will not be the eloquence of their labels. It will be whether, when a stranger opened the tab, there was something there to see.

So the next time a story tells you the field has been renamed, ask the only question that has ever mattered. Not whether it is true. Ask whether it is checkable. And then go check. The tab is one click away, and the truth — unlike the narrative — does not care whether you believe it.