David Marete

Aspiring Henokien. Pun intended.
Redeemed sinner. Uncle. Towards antifragile. SNR. Mitochondriac. Puns unintended.

[David Marete @dmarete]

Welcome to the fax machine.

[David Marete @dmarete]

2. From Demis Hassabis himself.
https://t.co/kwtbnZ4XP5

[Ihtesham Ali @ihtesham2005]

3/ Move 37 and the moment AI became creative

In 2016, AlphaGo played Lee Sedol in a match watched by 200 million people.

In game two it played Move 37. A stone on the fifth line early in the game. Every Go master alive would have called it a bad move.

Lee Sedol stood up and left the room. He needed 15 minutes to recover.

Move 37 won the game 100 moves later. AlphaGo had found something no human had found in 2,500 years of the game being played.

Hassabis said that was the signal he had been waiting for. The day they got back from Seoul he started AlphaFold.

[David Marete @dmarete]

1.

[David Marete @dmarete]

The AlphaGo move in question was described by a Go master who was watching the game as beautiful.
That was the point we should have turned everything off and gone back to rock art, hunting and gathering.

[David Marete @dmarete]

@SonOfAnchis3s 😅😅😅

[Dr. Shashank Joshi @AskDrShashank]

Never Lose Hope - Amazing story
A 24-year-old Polish tennis player arrived in Paris last week ranked 114th in the world, with no sponsors, no guaranteed income, and no certainty she could even pay for her hotel room.

She had to win three qualifying matches just to enter the French Open main draw. Prize money is only paid at the end of the tournament, so a Polish sports drink brand quietly stepped in and covered her hotel bill.

Her name is Maja Chwalinska. And today, she plays in the French Open final.

Before this tournament, she had won exactly one Grand Slam main draw match in her entire career. She had battled depression so severe that in 2021 she couldn't get out of bed. She underwent knee surgery in 2022. She spent years grinding through small tournaments across Europe just to stay afloat.

Then she arrived in Paris, won three qualifiers, and kept winning. Zheng Qinwen. Elise Mertens. Maria Sakkari. Diana Shnaider. Nine straight matches. One set dropped.

She is now the first qualifier in French Open history to reach the final. The last time a qualifier reached a Grand Slam final, it was Emma Raducanu at the 2021 US Open. Raducanu won.

By simply making the final, Chwalinska has earned more prize money than her entire career combined. The runner-up cheque alone is $1.6 million. If she wins today, she takes home $3.25 million.
One week ago she couldn't pay for her hotel room.

[David Marete @dmarete]

@BasedBiohacker BPC-157, lots of sunshine, red light therapy, an ankle support for the first week while cross training on the other ankle. Collagen, bone broth (Kobe did this after his sprain). Slow walking from week 2 with progressive overload.

[David Marete @dmarete]

@photobiogenesis Congratulations Champ!
A bit late in the day but this link should help https://t.co/okudYTKaH6
All the best!

[vittorio @IterIntellectus]

https://t.co/Qt46TOy2hT

[David Marete @dmarete]

While this is true, we will still need a moratorium on LinkedIn pieces when Arsenal wins the Premier/Champions leagues. 😅

[David Marete @dmarete]

All this rioting and EPL is not even over yet

[David Marete @dmarete]

The tongue of a blue whale weighs more than an adult elephant.

[David Marete @dmarete]

The AlphaGo move in question was described by a Go master who was watching the game as beautiful.
That was the point we should have turned everything off and gone back to rock art, hunting and gathering.

[Ricardo @Ric_RTP]

This guy just left Google and made a billion dollar bet that every AI company alive is building the WRONG technology.

He spent 13 years at Google DeepMind building the most legendary AI systems ever created.

AlphaGo: Beat the world champion at Go, a game experts said computers wouldn't crack for another decade.

AlphaZero: Taught itself chess from SCRATCH with just the rules. Beat the world's strongest chess engine within hours.

AlphaStar: Reached grandmaster level in StarCraft II against professional players.

And every single one learned the same way: NOT from human data.

They learned from their own experience. Playing against themselves millions of times until they discovered strategies no human had ever conceived.

That's the detail everyone's missing according to David Silver.

Because right now, the entire AI industry is doing the OPPOSITE.

ChatGPT, Claude, Gemini, and Grok all work the same way:

They inhale the entire internet, every book, every article, every Reddit post, and then predict the next word in a sequence.

That's the whole trick.

Silver says this approach has a hard ceiling.

These models can remix human knowledge. Summarize, rephrase, recombine what humans already wrote.

But they CANNOT discover something genuinely new.

They can't reason from first principles, plan 20 moves ahead, or understand physics.

A two year old knows what happens when you push a glass off a table. GPT-5 does not.

That's why AI hallucinates. No understanding of reality. Just word prediction.

So Silver quit DeepMind. Walked away from 13 years at the most prestigious AI lab on earth.

And yesterday raised $1.1 BILLION in a seed round.

It's the largest seed round in European history. A $5.1 billion valuation before even building a SINGLE product.

- Sequoia flew to London personally to lead the deal
- Nvidia wrote a check for at least $250 million
- Google invested
- The British government invested through their Sovereign AI fund

His company is called Ineffable Intelligence.

Its stated mission: "To make first contact with superintelligence."

This is not another chatbot or a coding assistant.

He's building what he calls a "superlearner." An AI that discovers ALL knowledge from its own experience without pre-training on human data, internet scraping, or copying what humans already know.

The same approach that produced the legendary Move 37 during AlphaGo's match against world champion Lee Sedol.

The AI made a move that had a 1 in 10,000 chance of being played by any human in history.

Every expert thought it was a bug. The live commentator said "I don't understand." Lee Sedol left the room for 12 minutes trying to process it.

It turned out to be the most brilliant move of the entire match.

A machine discovered something that 5,000 years of human Go mastery never found.

THAT is Silver's bet. Not AI that copies humans better but AI that thinks in ways humans literally cannot.

His company compared its own ambition to Darwin: "Where Darwin's law explained all Life, our law will explain and build all Intelligence."

The crazy thing is that this is the THIRD billion dollar raise by an AI researcher in two months.

Yann LeCun's AMI Labs raised $1 billion last month at $3.5 billion.

Mira Murati's Thinking Machines Lab raised $2 billion at $12 billion.

Now Silver at $1.1 billion and $5.1 billion.

3 researchers and 3 startups with zero products, revenue, or customers.

Over $4 BILLION raised on resumes alone.

Either these people are about to crack the most important scientific problem in human history.

Or the investors can't tell the difference between genius and hype.

What do you think?

[Dan Burmawi @DanBurmawy]

Iran struck the UAE more than any other state because the Islamic Republic understands exactly what the UAE has become.

Tehran was not attacking a neighboring Arab country that happened to host American bases. It was attacking the extension of Western civilization in the Arabian desert, the outpost of the Western order it has defined itself against for forty years, the living proof that an Arab Muslim state could choose modernity, openness, integration with Israel, partnership with Washington, and the entire civilizational package that the Islamic Republic’s founding ideology exists to destroy.

[How To Prompt @HowToPrompt__]

MIT proved every major AI model is secretly converging on the same "brain."

It’s called the “platonic representation hypothesis,” and it’s one of the most mind-blowing papers you’ll ever read.

You train a vision model purely on images. You train a language model purely on text.

They use completely different architectures. They process completely different data. They should have completely different "brains."

But as these models scale up, something impossible is happening.

When researchers measure how they organize information, the mathematical geometry is identical.

A model that only "sees" images and a model that only "reads" text are measuring the distance between concepts in the exact same way.

The models are converging.

The researchers named this after Plato’s Allegory of the Cave.

Plato believed that everything we experience is just a shadow of a deeper, hidden, perfect reality.

The paper argues that AI models are doing the exact same thing.

They are looking at the different "shadows" of human data, text, images, audio. And they are independently discovering the exact same underlying structure of the universe to make sense of it.

It doesn't matter what company built the AI.

It doesn't matter what data it was trained on.

As models get larger, they stop memorizing their specific tasks. They are forced to build a statistical model of reality itself.

And there is only one reality to map.

2024, Arxiv

[Rami @RamiSafadi93]

Muawiyah Siyasneh present at the trial of Atif Najib, the governor of Daraa who had him arrested and tortured for painting anti-Assad graffiti on his school when he was 15 years old. That was what sparked the entire revolution against Assad.

[Sid @AgarthaSid]

The least athletic version of LeBron James at 41 hanging with one of the most athletic players alive in Amen Thompson, throwing passes to his own son, and being among the most dominant players in transition is a sight to behold.

The ultimate athlete. And best QB on the floor.

[Yoko @Kibet_bull]

The trio that we want.

[David Marete @dmarete]

Other data point for boots on the ground pessimism.
If you check the wikipedia article on this action, American soldiers advanced a grand total of 8km from 8th April to 21st June, at this expenditure of life and limb.

[Matt Bracken @Matt_Bracken48]

These people are high on crack.
Qeshm Island is bigger than Okinawa, which took 3 months, 184,000 troops, and 12,500 KIA to take. And Queshm is in fire control range of countless Iranian missiles and drones dug into cliffs and caves for hundreds of miles back from Queshm on a 270* arc.

[David Marete @dmarete]

Correction: Russia has been transferring this knowledge to Iran. Not Ukraine.

[David Marete @dmarete]

If you are (still) following the Russia-Ukraine conflict, then you will have heard that the Russian spring offensive has collapsed and Ukraine now has the initiative. This has a major bearing on the war in Iran. Why?

[David Marete @dmarete]

As the good book asks in Luke 14:31
Or what king, going to make war against another king, sitteth not down first, and consulteth whether he be able with ten thousand to meet him that cometh against him with twenty thousand?