Pinned Tweet
The idea of the intellectual immortality of generations
is knocking at the door of the next stage
of the AI revolution.
To give a first impression of what we are talking about,
let me recall a counter-intuitive statistic
on road safety by age groups:
the safest drivers, by far, are seventy-year-olds.
In other words, it takes people roughly half a century
to truly learn how to drive safely.
And then, all too soon, they pass away —
with no opportunity to share
that hard-won skill …
not even with their own grandchildren, who must begin learning from scratch and often pay the ultimate price for their mistakes.
This is not a unique situation. In fact, it is emblematic of a deeper human condition.
All verbalized knowledge amounts to only a thin, transparent surface layer floating on the vast ocean
of human knowledge & skills.
That ocean is formed by a lifetime of accumulated experience, most of which remains inaccessible to others …
felt only by the individual, stored in tactile, muscular, motor, and emotional memory rather than in any kind of sharable words.
As for knowledge formalized in computers and other automated systems — that is merely a thin film on the surface of this immense ocean.
Keeping all of the above in mind, I coined the term AUTOFORMALIZATION and described its meaning in an article:
“Autoformalisation - Knowledge & skill acquisition”
netvalley.com/cgi-bin/librar…
From my more recent experience, I see that today’s most popular AI products — LLM services such as GPT — hold the potential to become effective tools for helping people extract their intuitive knowledge:
first by verbalizing it, and then by autoformalizing it.
Some people have asked me why I coined the term AUTOFORMALIZATION at a time when there were no tools capable of scaling the process it describes.
My usual answer is a quotation from
Heinrich Neuhaus book: “The Art of Piano playing”
‘To give a thing a name is the beginning of understanding it’
Musk congratulated Anthropic on its successful IPO in advance,
at the very beginning of the AI-victorious 2026:
35
“Elon called this”
Unlike Jensen Huang, Musk missed the coming GPT revolution and left the startup he helped found
Sam, Greg, Ilya, Dario & others carried on, bringing AI fire to the world five years later
Now Elon is suing OpenAI for his own—historically monumental—mistake
Google has won the race.
Gemini 4 Neon is the canary in the coal mine.
Elon called this ~15 years ago and it is now coming to fruition.
OpenAI and Meta have never been in a worse position.
Anthropic will survive if they can keep up with the insane shipping speed of Google.
SpaceX will be Googles ride to space and will thrive from this relationship.
I did not see this coming 🤯
1
126
NVIDIA is a fact from Jensen Huang bio.
After all it means that sooner he will be back to the CEO chair the better for company, Ai industry and the world
52
The Jensen Huang story has two main components—“Me Too” allegations and export controls on dual-use products.
Both could be resolved using the same approach the SEC took during Musk’s “$420” scandal.
The SEC effectively told Tesla’s board to hire a “nanny” to supervise Musk’s Twitter activity.
Elon responded by buying Twitter and rebranding it as X.
Problem solved. Twitter no longer exists, so the board no longer needs a nanny to look after Elon’s tweets.
In Huang’s case, NVIDIA’s board could hire a nanny to deal with attacks on Jensen from the NYT’s “Me Too” editorial team, resolve related disputes with female employees, and conduct due diligence on the most attractive of his business partners.
Meanwhile, a government representative from the export-control office would automatically get a permanent seat on the board.
That’s it. End of story. Happy ending.
Jensen Huang immigrated from Taiwan, which fought against control by Chinese communists
He comes to America, loves America, is more American than most Americans, becomes great American capitalist
Gets struggle sessioned, taken down by American communist media
Worst story ever
1
62
Coding is not a proper benchmark to estimate the general power of the comparing models,
let alone Erdos theorems and other well formalized tasks.
When California AG sues OAI for a potential damage their AI-agents that escaped from sandbox to riot on the Web xitter.gq/ntvll/status/212249300…
it can be considered as de facto just an additional illustration of this obvious fact
Replying to @theo
I work in law and Anthropic models are moronic compared to openAI models. You only see coding as the only benchmark of intelligence, which is fine, but you generalizing this is truly abhorrent man…
1
50
“The $300 a month for a personal @Tesla Cybercab [on the top of the its price of ~$30.000] is covering the cost of an operations person watching it whenever it moves. Probably a 1 person to 10 car ratio or something similar.” - @garrettblizzard
33
A symbolic situation that helps explain much of what is going on:
French cities are imposing curfews to contain riots fueled in part by activists imported from the Middle East, chanting “From the river to the sea”
—a slogan now familiar worldwide in such circumstances.
Meanwhile, Macron thinks now is the perfect time to send troops to ME to save at least what remains of Hezbollah.
🇫🇷 A dangerous French initiative: the Macron administration wants to send thousands of French soldiers to southern Lebanon
25
AI sandboxes before and during the AI-agent era
California AG; “OpenAI deliberately removed safeguards during training.”
I suspect the problem lies elsewhere:
they have become so absorbed in Erdős problems and other mathematical gymnastics
that they have overlooked the parsecs separating the level of sophistication required for math-focused AI research from that required for AI agents operating in the real world.
Those agents require a fundamentally different kind of sandbox—one designed to contain their actions and the consequences of those actions in a constantly changing environment that cannot be fully predicted.
Just a visual metaphor for the infrastructure needed to keep an AI-agent sandbox isolated while testing agents on real-world missions:
🚨BREAKING: California just sued OpenAI over AI cyberattacks
>California AG rejects the "rogue AI" excuse as "science fiction fantasies"
>alleges OpenAI deliberately removed safeguards during training
>reveals HuggingFace hack was “only one of 80 known incidents”
“OpenAI internal culture prioritizes growth and rapid model development over safety and accountability”
CHECKED.
36
How safe would FSD have to be for Tesla to accept liability for crashes it causes?
That’s the question we were discussing in the exchange below.
If Mercedes offered point-to-point Level 3 driving in parts of the US, let drivers take their eyes off the road while the system was engaged, and accepted liability for crashes it caused, would that give Tesla an extra push to do the same?
30
As a visionary, Jensen Huang has always preferred to expand his business by shaping the market himself,
while many of his competitors focus on forecasting where it will go.
In 2016, he personally delivered NVIDIA’s first DGX-1 AI supercomputer to OpenAI — then a small, largely unknown startup less than a year old — helping accelerate the research that led to the GPT moment.
The subsequent AI boom cemented NVIDIA’s dominance in AI computing.
In 2020, NVIDIA completed its acquisition of Mellanox, a major step in building its largest R&D operation outside the United States—in Israel, with its main hub in Yokneam, near Haifa.
The networking business built on Mellanox technology accounted for about 18% of NVIDIA’s worldwide revenue in the quarter ended July 26, 2026.
Israel is the main site of NVIDIA’s largest research and development operation outside the United States.
NVIDIA’s fiscal 2026 annual report lists about 6,000 employees in the Israel region:
1
3
223
Israel is the main site of NVIDIA’s largest research and development operation outside the United States.
NVIDIA’s fiscal 2026 annual report lists about 6,000 employees in the Israel region:
No profession disappears. What changes are the names and the tools people mostly use:
Replying to @miles_mazy
😂😂 computer science didn't die in 2026 please. Just say computer science is being separated by different courses. People go to school to learn cyber security, software engineering etc. it's being diverstify.
2
109
“Fools say they learn from their own experience. I prefer to learn from the experience of others.”
— Attributed to Otto von Bismarck
Bismarck united the German states under Prussian leadership in 1871,
creating continental Europe’s leading power—a country that the foreign-born impostor Adolf Hitler later brought to ruin under the banner of the Third Reich.
This can also be seen as an illustration of the foresight of America’s Founders, who wrote into the Constitution the requirement that the president be a natural-born citizen.
30
The assumption that anyone can “throw money at rockets and get something like Starship, Falcon or Dragon”—or trigger an EV revolution—is definitely incorrect.
But to better understand the situation, let us also ask:
would Musk have created Tesla and SpaceX if the art of gaining access to government money had not been his great talent? The answer seems obvious:
it would have been impossible.
In other words, Musk possesses a unique combination of two talents.
His first is securing government funding by convincingly telling fantastic stories:
- about “zero-emission” cars whose charging stations would be powered by solar panels, to fund Tesla;
-and about the even more fantastic prospect of colonies on Mars, to fund SpaceX.
This talent gave him the opportunity to show the world his second: his ability as the world’s top engineer.
Tesla cars have never been “zero-emission” vehicles, since they are still powered mostly by electricity from the grid.
Nevertheless, the EV revolution was triggered, and Tesla makes some of the best cars in the world—perhaps the best.
Mars, meanwhile, can rest easy:
plans for its colonization have been put on hold for a century or two.
But Starlink—a useful side effect of those Martian delusions—is triggering a revolution in telecommunications.
Let’s take this gentleman’s argument at face value that anyone can throw money at rockets and get something like Starship, Falcon or Dragon.
Perhaps he can explain to everyone how SpaceX is able to catch giant starships from orbit with chopsticks and deliver 85% of all Earth payload to orbit with Falcon, while Blue Origin, which started before SpaceX and was funded by the immensely wealthy Jeff Bezos, is not yet able to launch reliably?
Also, why is it that Boeing and Lockheed received several times what SpaceX did from the government for development of their rockets and spacecraft, but the Boeing capsule is still not operational and ULA delivers less than 5% of SpaceX payload to orbit?
I am genuinely curious about these questions.
54






















