Kurzweil Scorecard: The Universe Didn’t Wake Up. The Copy Did.

There is a patent, granted in September 2025, that describes training a language model to keep posting after you die.

US 12,513,102 — “Simulation of a user of a social networking system using a language model” — is written in the flat prose of a filing, but read the description and the temperature drops. The model is trained on a person’s own interaction history. It reads their feed, generates the reply “the target user would have performed upon viewing the content item,” and posts it under their name. The stated use case is not coyly hidden: the system exists to cover a user who “is absent from the social networking platform,” and the impact “is much more severe and permanent if that user is deceased and can never return.” One dependent claim provides for “a plurality of language models trained to simulate the target user at a particular age” — several copies of you, indexed by how old you were. Another asks the still-living original for permission first.

This is the closing chapter of Ray Kurzweil’s The Singularity Is Near — the one reviewers filed under mysticism — rendered as a claims table. It is the last real batch in this scorecard series, and it is the one where prediction and philosophy stop being separable. So let’s take the metaphysics seriously, and see what reality actually did with it.

The predictions

In the book’s final movement, Kurzweil stops forecasting gadgets and starts forecasting selfhood. He argued that personal identity is best understood primarily as a pattern of matter and energy persisting over time rather than as a fixed set of particles (ch. “Who Am I? What Am I?”) — a position he calls patternism. From that he draws a chain: that gradual replacement of brain parts with neuromorphic equivalents appears to preserve consciousness and identity continuously even as it raises paradoxes about whether the original self survives; that a copied mind behaviorally identical to the original would still not be the same person if the original remains in existence; and, at the largest scale, that once intelligence saturates the matter and energy of the universe, the universe will wake up and become sublimely intelligent in Epoch Six (ch. “Ich bin ein Singularitarian”).

In The Singularity Is Nearer (2024), he restates patternism without flinching. The self, he writes, is like a river: “even though all of the water molecules passing through a given slice of it are completely different every few milliseconds, those molecules act in a consistent pattern that defines the course of the river. It is the same for minds.” He grounds it in cell biology — half your neuronal mitochondria turn over in a month, NMDA receptors in synapses are replaced in hours, dendritic actin filaments last about forty seconds — so “your brain is almost completely replaced within a few months.” You are already a copy of yourself. And on the Sixth Epoch he is unchanged: “The Sixth Epoch is where our intelligence spreads throughout the universe, turning ordinary matter into computronium, which is matter organized at the ultimate density of computation.”

Twenty years on, two of these have quietly become infrastructure. The other set can’t be scored at all — and Kurzweil, to his credit, told us why.

Where we actually are

Patternism became load-bearing before anyone proved it. The idea that a self is a substrate-independent pattern is no longer a debate topic; it is the founding assumption of an industry. Work on whole-brain emulation, mind uploading, and connectomes climbed from a few dozen substantive papers a year around 2010 to more than 400 a year by 2020. The empirical spine of that field arrived in October 2024, when the FlyWire consortium — 127 institutions — published the first complete wiring diagram of an adult Drosophila brain in Nature: 139,255 neurons and over 50 million synapses, the most complex brain ever fully mapped. That is patternism’s proof of concept — a mind captured as a pattern of connections, in principle re-instantiable. It also shows how far the human version is: a fly’s 140,000 neurons against our 86 billion.

The substrate half of the “gradual replacement” prediction runs ahead of the fly map. Neuromorphic patents — chips that compute the way neurons do rather than the way processors do — have held at 60 to 80 grants a year since 2019. In April 2024, Intel switched on Hala Point at Sandia National Labs: 1,152 Loihi 2 chips wired into 1.15 billion neurons, roughly the neuron count of an owl’s brain or a capuchin monkey’s cortex, drawing 2,600 watts. That is Kurzweil’s “neuromorphic equivalent,” at monkey-cortex scale, four years ago. And brain-computer interface and neural-prosthesis patents — the machinery of swapping biological parts for engineered ones — have climbed from single digits a decade ago to 38 grants in 2025. Kurzweil’s own example, a prosthesis doing the hippocampus’s work, is now in human trials, not a thought experiment.

The copy problem stopped being philosophy and became product design. Kurzweil’s cleaner claim — that a perfect behavioral copy is not you if you’re still around — is precisely what US 12,513,102 has to legislate. When a patent has to distinguish the copy that posts while you’re on vacation from the copy that posts after you’re gone, and build a permissions layer so the living original can consent to being simulated, the “am I the same person” question has been converted into a settings menu. Patents describing digital twins, avatars, or chatbots tied to a specific person’s personality or memory rose steadily from about 30 grants in 2018 to 93 in 2025. The market answered Kurzweil’s thought experiment the way markets answer everything: by shipping it and letting the courts sort out the metaphysics later.

But the universe did not wake up, and — this is the real finding — we could not measure it if it had. The grand predictions rest on a hidden assumption: that consciousness is something we can detect and quantify. In 2005 that looked like a matter of time. It is not going well. In 2023, 124 researchers signed an open letter branding Integrated Information Theory — the leading formal, mathematical theory of consciousness — “pseudoscience,” largely because its logic points toward panpsychism, the view that consciousness is fundamental and everywhere. That is not a fringe collision; it is exactly the worldview Kurzweil edges toward when he flirts with panprotopsychism in the 2024 book. The field’s own flagship experiment underscores the impasse. The Cogitate consortium ran a preregistered adversarial collaboration — theory proponents agreeing in advance on what would falsify their views — pitting Integrated Information Theory against Global Neuronal Workspace theory across 256 subjects on fMRI, MEG, and intracranial recordings. The result, published in Nature in 2025, was a split decision: posterior cortex looked sufficient for conscious content, denting the workspace theory, but neither framework emerged intact. After a preregistered showdown with the best tools we have, we still cannot say which theory of consciousness is right — for a human, wide awake, looking at a picture.

Against that, “the universe will become conscious in Epoch Six” is not wrong. It is unrankable. Kurzweil is honest about this in a way his critics rarely credit. On uploading a mind, he writes that whether the copy is truly you is “impossible to prove or disprove scientifically, and each of us will have to make decisions about using this technology based on our own philosophical or spiritual values.” That sentence is the whole scorecard for this batch. The engineering is proceeding as if patternism were settled. The science that would settle it cannot yet measure the thing in question.

The scorecard

What Kurzweil missed (and what he nailed)

The pattern across this final batch is the mirror image of every batch before it. Elsewhere in this series, Kurzweil’s misses were misses of mechanism — right destination, wrong road: the atoms went programmable but not through Drexler’s assemblers, the brain got read but not decoded. Here the error runs the other way. The concrete, almost throwaway idea — you are a pattern, so a pattern can be you — spawned real patents, real chips, and a real market for reanimating the dead. The soaring idea — the cosmos becoming a single mind — turned out to be not premature but unmeasurable, blocked not by engineering but by the fact that we can’t yet define, detect, or agree on the thing it predicts.

Kurzweil’s most reliable instinct was never his optimism about dates. It was his willingness to state a mechanism concrete enough to be checked. When he said “pattern, not particles,” the patent office eventually had something to grant. When he said “the universe wakes up,” he wrote a sentence no experiment can reach — and, unusually for a futurist, said so. The copy will keep posting. Whether anyone is home inside it is a question our best instruments, twenty years on, still return blank.

Method note

This analysis draws on a full-text index of roughly 9.3 million U.S. patents and a corpus of about 357 million research papers, queried for filing and publication trends by year, plus the full text of specific granted patents read claim by claim. Consciousness-science findings, the fly connectome, and the neuromorphic hardware milestone were confirmed against primary sources published in Nature and by the institutions involved. Predictions are quoted from The Singularity Is Near (2005) and cross-checked against Kurzweil’s restatements in The Singularity Is Nearer (2024). Verdicts reflect the evidence as of mid-2026; philosophical predictions are marked unfalsifiable where no empirical test exists, rather than forced into a score.