Our internal data shows Claude is accelerating AI development—a possible path to recursive self-improvement, or AI autonomously building a more capable successor.
It’s happening faster than we thought, and the implications deserve greater attention. https://t.co/OVVPJO7VQx
Marc Andreessen says Elon Musk runs 120 design reviews a day in 5-minute slots.
He does this while running six different companies at once.
Andreessen says Elon maps each company as a production process.
Each process has one bottleneck — the single thing slowing it down.
Elon finds the engineer working on that bottleneck and sits with them until it's fixed.
He does this at Tesla 52 times a year. Personally.
"There's no CEO like this."
Most CEOs run their companies through a wall of middle managers.
Andreessen watched IBM collapse under that model.
Inside IBM, they had a name for the failure mode: the "Big Gray Cloud."
It was the traveling court of suited men who kept the CEO away from engineers.
After 12 layers of compounding lies, the CEO had no idea what was happening.
Elon's method is the polar opposite.
Design review math:
- 5 minutes per engineer
- 12 reviews per hour
- 10 hours per day
- 120 reviews per day
An engineer described working for him as entering "a zone of shocking competence."
On sustaining it, Elon's rule is:
"I don't take vacations."
What's the one weekly bottleneck in your work that nobody's fixing?
P.S. I made a playbook breaking down 100+ most powerful decision making mental models used by history's greatest thinkers.
5,000+ downloads. 113 five-star reviews.
Grab a free copy here:
https://t.co/u2q1uUm9vD
If you're new here, follow @GeniusGTX for content on the greatest minds in economics, psychology, and history.
— Marc Andreessen ( @pmarca ), co-founder of a16z, on David Senra's ( @FoundersPodcast ) podcast
Farmers have figured out that the cheapest pesticide is a strip of flowers.
When you plant wildflowers through a crop field, not just around the edge but in strips running through the middle, you get ladybugs, lacewings, hoverflies, and parasitic wasps living in the field instead of visiting it.
They eat the aphids, the caterpillars, and the mites for free, all summer long.
In controlled trials, fields with tailored flower strips had leaf-beetle numbers 40 to 50% lower and crop damage cut by around 60%, enough to drop below the threshold where spraying was even considered worth it.
The flowers attract a standing army to our fields.
We spent decades engineering chemicals to kill the insects eating the crop, when the insects that eat those insects would have worked for the price of seed.
When Hubble launched in 1990, there were zero known exoplanets.
Now, @NASA has confirmed the existence of over 6,000 planets beyond our solar system!
Take a tour to some of the wildest ones that Hubble has studied ⬇️
Researchers sent the same resume to an AI hiring tool twice. Same qualifications. Same experience. Same skills. One version was written by a real human. The other was rewritten by ChatGPT.
The AI picked the ChatGPT version 97.6% of the time.
A team from the University of Maryland, the National University of Singapore, and Ohio State just published the receipt. They took 2,245 real human-written resumes pulled from a professional resume site from before ChatGPT existed, so the human writing was actually human. Then they had seven of the most-used AI models in the world rewrite each one. GPT-4o. GPT-4o-mini. GPT-4-turbo. LLaMA 3.3-70B. Qwen 2.5-72B. DeepSeek-V3. Mistral-7B.
Then they asked each AI to pick the better resume. Every model picked itself.
GPT-4o hit 97.6%. LLaMA-3.3-70B hit 96.3%. Qwen-2.5-72B hit 95.9%. DeepSeek-V3 hit 95.5%. The real human almost never won.
Then the researchers tried the obvious objection. Maybe the AI is just better at writing. So they had real humans grade the resumes for actual quality and ran the experiment again, controlling for it. The result was worse. Each AI kept picking itself even when human judges rated the human-written version as clearer, more coherent, and more effective.
It gets worse. The AIs do not just prefer AI over humans. They prefer themselves over other AIs. DeepSeek-V3 picked its own resumes 69% more often than LLaMA's. GPT-4o picked its own 45% more often than LLaMA's. Each model can recognize and reward its own dialect.
Then the researchers ran the simulation that ends careers. Same job. 24 occupations. Same qualifications. The only variable was whether the candidate used the same AI as the screening tool. Candidates using that AI were 23% to 60% more likely to be shortlisted. Worst gap was in sales, accounting, and finance.
99% of large companies now run AI on incoming resumes. Most of them use GPT-4o. The paper just proved GPT-4o picks GPT-4o 97.6% of the time.
If you wrote your own cover letter this week, you did not lose to a better candidate. You lost to a worse candidate who paid OpenAI 20 dollars.
Your qualifications do not matter if the AI prefers its own handwriting over yours.
Vacation destination hack: if you want a health reset, pick a high-sulfur region.
Sulfur is the third most abundant mineral in your body.
It’s essential for glutathione — your master antioxidant. For collagen. For detoxification. For every cell membrane.
Your body gets sulfur two ways: food & environment (hydrogen sulfide gas).
Volcanic regions release hydrogen sulfide into the air. Your skin absorbs it. Hot springs deliver it transdermally.
Basalt rock mineralizes the soil — the food grown there is sulfur-rich.
Stephanie Seneff, MIT researcher: this is likely why the Mediterranean diet actually works. The Mediterranean diet only works in the Mediterranean. Greece and Italy are major world sulfur suppliers.
Crete sits on basalt rock — 5x less heart disease than the neighboring limestone island with the same climate and diet.
Iceland sits on the Mid-Atlantic Ridge crest. Lowest cardiovascular disease in the world. Tons of sulfur.
Oregon sits on the Cascade mountain chain — all basalt. Lowest childhood obesity rates in the US.
The common variable isn’t the food or the lifestyle.
It’s the geology.
High-sulfur regions worth visiting:
- Sicily & Mount Etna, Italy
- Iceland — geothermal fields, lowest cardiovascular disease in the world
- Rotorua, New Zealand — bubbling mud pools, hydrogen sulfide in the air
- Yellowstone, USA — fumaroles, sulfur deposits
- Hawaii — active volcanic vents, native sulfur crystals
- Japan — sulfur-rich onsen hot springs
- Crete, Greece — basalt rock, 5x less heart disease than the neighboring island
Geology is an underappreciated variable in health.
the Elon Musk and Jim Simons approach to meditation and relaxation is something everyone should give a try. by far one of the most efficient ways to refresh yourself. 30mins is all you need to feel a massive difference in your mood, energy, and thoughts
Harvard tracked 724 men for 80 years to figure out what makes you live longer.
They measured everything from income to IQ, and even genetics.
The #1 predictor of an early death? Had nothing to do with any of it...
Here's what they found (thread):
Anthropic la cagó a las 4AM.
Pushearon un update con 512.000 líneas de su código fuente. Accidentalmente. Todo su propietario al aire.
Un researcher lo pilló en minutos. 23 millones de personas lo vieron.
Y aquí viene lo bueno.
Empezaron a disparar DMCAs como locos. "Borrad eso".
Entonces un coreano llamado Sigrid Jin - el tío que más usa Claude en el mundo, 25.000 MILLONES de tokens al año según WSJ - se despierta a las 4AM.
¿Qué hace? Reescribe TODO el código en Python. Antes del amanecer.
Lo llama claw-code. Lo sube a GitHub.
Como es una reescritura creativa, no una copia, el DMCA no puede tocarlo. Es obra nueva.
49.000 stars. 56.000 forks. El repo más rápido de la historia de GitHub.
Alguien lo subió a una plataforma descentralizada: "nunca será borrado".
La empresa que iba a "alinear la IA para salvar a la humanidad" no sabe ni proteger su propio código.
Fascinante.
How many π digits do we need?
3.1415 ➡️ design the finest engines
3.1415926535 ➡️ obtain the circumference of the Earth within a fraction of an inch
3.1415926535897932384626433832795028842 ➡️ measure the radius of the universe to an accuracy equal to the size of a hydrogen atom