Loading…

AI‑IFICATION

The Convergence Has Begun.

Biology Computation Energy The Human Mind

— Artificial Intelligence —

Scroll

For most of history, breakthroughs happened inside separate disciplines — biology stayed biology, physics stayed physics. That's changing. AI is becoming the connective tissue between genetic engineering, quantum computing, fusion energy, and the human brain — not a headline in any one of them, but the thing increasingly present inside all four.

The Window

Everything In The History Of Humanity That Has Never Happened Before Happens Here.

If humanity lasts as long as the Sun allows, about 5 billion more years, this is where every “first” sits. It may not be a coincidence. Why now? ↓

N = R* · fp · ne · fl · fi · fc · L

The Drake equation estimates how many civilizations are out there. The L factor at the end is how long a civilization lasts.
We may not be living in a random moment. We may be living in the moment when the “firsts” happen. Perhaps L is short.

Dawn of human culture
Today (our 500 year epoch)

Every “first” on this page — all inside the 500-year epoch we are living through, 0.00001% of the road ahead. Told as if looking back from the end of it. Hover or tap one.

None of these things can ever happen for the first time again.

Today +1 billion years +2 billion years +3 billion years +4 billion years +5 billion years Possible future of humanity: up to ~5,000,000,000 years → Oceans likely lost to a brightening Sun (~1–1.5 B yrs) The Sun exhausts its hydrogen and swells to engulf Earth ~5 billion years from now
×1,000 magnification Window: from the dawn of culture to 2,800,000 years ahead Today +1,000,000 years +2,000,000 Dawn of human culture cave art, ~30,000 years ago Today — every “first” on this page: the 500-year epoch we are living through (the atom bomb to today is only 81 years — invisible at this scale) the road ahead: 5 billion years, far off the right edge → Even at 1,000×, 30,000 years of culture and every first are one mark.

Drawn to true scale, except that the spike is widened so it can be seen. At real scale it would be thinner than a pixel. The graph below magnifies the area around Today 1,000 times.

The Question Everyone Avoids

At What Step Does It Become Conscious?

Drag the slider from a patch of light-sensitive skin to a man crying at a concert. At each stop, answer honestly: is this conscious yet? There are no wrong answers, but you will have to pick a place.

The eye
The network behind it

A patch of skinA man crying at a concert
Is this conscious?
Where readers say the lights come on
YesNot sureNo

The ladder needs no new ingredient at any rung: each step is the one before with more connections, more memory and more loops. It doesn’t show that a blink feels like anything, or that the man’s grief is an illusion. It shows that wherever you put the line, you have to say what physically changed there.

2023
First CRISPR-Based Cure Approved (Casgevy, gene HBB)
Verified
FDA, 2023
Dec 2024
First Below-Threshold Quantum Error Correction (Google Willow)
Verified
Google, Dec 2024
75%
SPARC Fusion Reactor Construction Complete
Engineering Target
Commonwealth Fusion Systems, Feb 2026
31
Human Patients Implanted (Neuralink + Synchron)
Clinical
Company disclosures, early 2026
~2030
Demis Hassabis's ~50% Odds on Human-Level AGI
Forecast
Davos, Jan 2026
Four Frontiers

Where AI Is Already Doing the Work

Stage 1

AI Rewrites the Genome

CRISPR-Cas9 is a programmable molecular scalpel borrowed from a bacterial immune system: a short guide RNA finds a matching 20-letter sequence anywhere in the genome, and the Cas9 enzyme cuts both DNA strands exactly there. The cell's own repair machinery does the rest — disable a gene, correct a single letter of DNA, or, with newer base and prime editors, rewrite it without cutting at all.

20bp
Guide RNA Match
Lab
NGG
Cas9 PAM Site
Verified
2012
CRISPR-Cas9 Debut
Verified
2023
First CRISPR Cure
Verified

Not a mockup: this is the actual T2T-CHM13v2.0 sequence, the first-ever gap-free human genome, streamed base-by-base from this server as you navigate.

chr11 · ~1.6 kb visible Live T2T-CHM13v2.0
🧬 Gene of the Week: HBB — first condition ever cured by CRISPR (2023)
Or jump to a famous gene:
A C G T Gene
57,000+ real NCBI RefSeq gene coordinates. Zoom in past a few hundred bases to see actual A/C/G/T letters (or search chr1:1,000,000-1,000,050).
Full deep-dive →
Stage 2

Quantum Computing, Decoded by AI

Quantum computers are still fragile — a qubit's state collapses if you so much as look at it wrong. AI is what's making them practical anyway: machine-learned decoders now read a chip's error syndromes and correct them faster than any hand-written algorithm, while the long-term bet runs the other way — a fault-tolerant quantum computer could eventually hand AI a kind of computational shortcut nothing classical will ever match.

Demonstrated: below-threshold error correction, 84-1,000+ qubit chips. Speculative: a fault-tolerant machine actually accelerating AI itself.

4,158
IBM Kookaburra Qubits
Engineering Target
50
Quantinuum Logical Qubits
Verified
64 AQ
IonQ Tempo
Verified
10×
Faster AI Error Decoding
Verified
Google
Willow achieved the first below-threshold quantum error correction; DeepMind's AlphaQubit model now reads its error syndromes and decodes them faster and more accurately than classical decoders.
IBM
Kookaburra, a ~4,158-qubit system due by end of 2026, targets a real quantum-advantage workload; an AI-driven decoder already cut error-correction time 10×, a year ahead of IBM's own roadmap.
Quantinuum
Helios ships 50 error-corrected logical qubits at 2:1 encoding — more than any other machine — on trapped-ion hardware AI helps calibrate shot by shot.
IonQ
Tempo reached 64 algorithmic qubits in early 2026 with all-to-all trapped-ion connectivity, a very different hardware bet than Google's and IBM's superconducting chips.
Full deep-dive →
Stage 3

Machine Intelligence Tames the Plasma

A fusion plasma is unstable on a millisecond timescale — too fast for a human operator to react to. Reinforcement-learning controllers now steer magnetic confinement and predict plasma-destroying instabilities before they happen, doing in software what used to take a room full of engineers watching dials in real time.

Demonstrated: AI-predicted instabilities, real magnet installs. Speculative: any commercial plant producing grid electricity within the next year.

20+ T
CFS Magnet Field
Verified
200ms
AI Instability Warning
Verified
2027
SPARC Net-Energy Target
Engineering Target
2028
Helion Grid-Power Target
Engineering Target
Commonwealth Fusion Systems
SPARC's final D-shaped magnet went in this year, generating a 20+ tesla field; a DeepMind-built AI pilot now steers the reactor's magnetic shape and heat load in real time, targeting net energy in 2026–2027.
Helion Energy
Backed by Sam Altman, targeting first grid electricity by 2028 under a 50MW deal with Microsoft; AI tunes each plasma pulse using data from the one before it.
TAE Technologies
A decade-long collaboration with Google, the "Optometrist Algorithm," pairs machine learning with human intuition to pick the next best plasma experiment out of a search space too large to brute-force.
DeepMind × EPFL
The research seed for all of it: a reinforcement-learning controller trained in simulation ran a real tokamak (TCV) end-to-end, then learned to flag disruptive instabilities roughly 200ms before they happen.
Full deep-dive →
The Convergence

These Curves Do Not Exist in Isolation.

Four fields, historically unrelated, now visibly pulling on each other through one connective technology. Scroll a little further and watch what happens.

Genomereads & writes DNA
Quantumfaster simulation
Fusionpower for compute
Neuralnew kind of data
AI
Convergence

The Flywheel: Toward Artificial Superintelligence

Each field above both depends on AI and feeds it back. A fault-tolerant quantum computer hands AI a computational shortcut nothing classical can match. A fusion reactor AI can keep stable indefinitely hands AI something to run on — functionally unlimited clean power for functionally unlimited compute. A genome AI can read and write, plus a neural interface AI can read and write to, close the loop on the last analog system left: us. Push all four far enough and the line between "a very good tool" and "a new kind of mind" stops being obvious. That's the case for artificial superintelligence — not one breakthrough, but four curves accelerating each other. None of this is guaranteed; what's real is that the feedback loops now exist.

Dario Amodei2026–2027ForecastAI "broadly better than all humans at almost all things," said at Davos, January 2026.
Sam Altman"A few thousand days"SpeculationHis own phrase for how far off superintelligence is — years, not decades.
Demis Hassabis~2030, ~50% oddsForecastPut a number on genuine human-level AGI at Davos 2026: five-to-ten years out, roughly a coin flip by decade's end.
Why Now?

Everything Is Happening For The First Time.

Look at a 2,000-year window, the thousand years behind us and the thousand ahead. Inside it, the most consequential events in the history of our species are landing together, in roughly the same few generations. Each of these can only happen for the first time once. A billion years from now, no one will map the Earth for the first time, or edit the human genome for the first time, or build the first mind that isn’t born. Humans may have billions of years ahead of us. So why are we here now?

All of the “firsts” now sit on the timeline at the top of the page. Hover any of them there for the full story.

Zoom In: The Spike

The 2,000-year window, from the year 1000 to the year 3000. Almost everything lands in the last 80 years.

1000 2000 3000 Today (2026) 1945–2030 1945 Trinity · 1961 first human in orbit · 1969 Moon + ARPANET 1980 smallpox eradicated · 2003 genome sequenced 2012 CRISPR + Voyager 1 leaves · 2022 fusion ignition + DART 2020s AI matches human language · superintelligence forecast Most of human history, flat ?? Unknown future

What this might mean for the Drake equation

N = R* · fp · ne · fl · fi · fc · L

L is how long a technological civilization lasts. If civilizations tended to survive for billions of years, then most of the observers who ever exist would live billions of years after their civilization’s own “firsts,” and finding yourself inside the first 2,000 years of a 5-billion-year run would be a roughly 1-in-2,500,000 coincidence.

One way to remove the coincidence is a short L. If civilizations like ours usually last only centuries or a few thousand years, then this window is not a tiny slice of a long story. It is most of the story. That would also explain the silence of the sky: at any moment there are very few civilizations to hear.

This is an argument, not evidence. It is the doomsday argument and the Fermi paradox pointing the same way, and it has serious critics, including those who argue that observer-selection reasoning is unreliable. The 5-billion-year figure is also an upper bound set by the Sun, not a prediction. The point of the timeline is to make the question visible.

Speculation

Why are we here, now?

Four serious answers, and we don’t know which is right.

The hinge of history

Philosophers Derek Parfit, Toby Ord and Will MacAskill argue this century may be the most pivotal ever. If we survive the dangerous stretch, we could last for billions of years.

Forecast

The observer effect

Any being that can ask this question must live after intelligence and technology arrive. Observers would find themselves near a transition like this one, whenever it happened, so it may be less of a coincidence than it feels.

Speculation

The doomsday argument

If you are a random person among all humans who will ever live, you are probably not among the very first. Taken at face value, that suggests a shorter future than billions of years.

Speculation

Maybe not a coincidence

Perhaps every intelligent species reaches this stage: it learns to rewrite life, build minds and harness stellar energy within a few generations. The silence of the universe could mean most don’t get through it.

Speculation
We may not be living in a random moment. We may be living in the moment when the “firsts” happen.
— AI-ification.AI
This Week in the Future

Five Things That Actually Happened

VerifiedFeb 2026 · Fusion

CFS installs SPARC's first 20+ tesla magnet

Crosses a major engineering threshold toward the 2026–2027 net-energy target.

VerifiedMar 2026 · Neural

Eon Systems runs a real fly connectome in physics simulation

A mapped biological brain's wiring alone — not training — produced working behavior in simulation.

Why this matters to AI →A real connectome running in simulation is a preview of what training on biological neural data, not text, could eventually look like.
Verified2026 · Quantum

NIST-administered $1.8B quantum funding round lands

National governments, not just venture capital, are now underwriting the qubit race.

Lab2026 · Genome

AlphaFold3 contact modeling improves base-editor selectivity

A protein-structure model built for basic biology just made a gene-editing tool better.

Why this matters to Human Augmentation →Editors that are better-targeted are editors that are safer to use outside of last-resort disease treatment.
ForecastJan 2026 · ASI

Dario Amodei predicts broadly superhuman AI by 2026–2027

Frontier-lab timelines for broad superhuman AI have compressed to single-digit years, not decades.

The Convergence · Weekly

Five Frontiers. One Week. The Connections That Matter.

Not another AI newsletter. Every week: what changed across genome editing, quantum computing, fusion, and brain interfaces — and, more importantly, how the changes connect.

Issue No. 9 · Nov 2, 2026

Doubled in Cells, Simulated in Silicon, Promised in a Post: Four Fields, Four Distances from Proof

A protein language model designed a DNA-insertion enzyme that doubled targeted insertion in cultured cells. NVIDIA simulated 104 fault-tolerant qubits with no AI in the description. Commerce signed a letter of intent for up to $85 million for Commonwealth Fusion electronics. And Musk says Neuralink's Blindsight is engineering-ready, with no data.

Covered this week: Genome, Quantum, Fusion, Neural