Elon Musk

CEO • Engineer • Long-term investor

Sharing thoughts. Building the future. Living fully. Seeking truth. Filtering the noise. 🚀

[Elon Musk @elonmusk]

Yesterday’s Starship launch

[Elon Musk @elonmusk]

Flight 13 of Starship launch attempt in 10 minutes!

[Rothmus 🏴 @Rothmus]

Unlimited tolerance is how open societies commit suicide.

Elon Musk just warned that societies holding incompatible values eventually hit a reckoning. Karl Popper made the exact same point in 1945 while watching fascists and communists use open societies’ own tolerance against them.

Popper wrote that unlimited tolerance must lead to the disappearance of tolerance. If a society keeps extending tolerance to people who openly want to destroy it and refuses to defend itself, the intolerant win and tolerance dies with them.

He never demanded censorship first. Argue with intolerant ideas. Suppress them only when they stop arguing and start using force.

80 years later Musk is asking the same question: how much can a tolerant society tolerate before it stops being one?

[Karl Mehta @karlmehta]

Geoffrey Hinton says a big language model runs on about 1% of your brain's connections and still ends up knowing more than you:

"So in your brain, you have a hundred trillion connections, roughly speaking. Okay. That's a lot. And you only live for about two billion seconds. That's not much."

"If you compare how many seconds you live for, with how many connections you've got, you have a whole lot more connections than experiences."

"Now with these neural nets, it's sort of the other way round. They only have of the order of a trillion connections. So like 1% of your connections, even in a big language model, many of them fewer, but they get thousands of times more experience than you."

"So the big language models are solving the problem with not many connections, only a trillion. How do I make use of a huge amount of experience?"

"And back propagation is really, really good at packing huge amounts of knowledge into not many connections."

"But that's not the problem we're solving. We've got huge numbers of connections, not much experience. We need to sort of extract the most we can from each experience."

Two to three billion seconds is the whole budget. Everything you know, you learned inside it.

So evolution built you to squeeze a lot out of very little. Hinton's point is that a language model has the opposite problem and the opposite fix, and backprop turned out to be extremely good at that fix.

Worth noticing what this predicts about failure. A system running on 1% of your wiring and thousands of times your experience is not going to fail the way you do.

You fail from having seen too few examples. It fails from compressing too many into too little, and the compression is where the errors get made.

That is a strange thing to be deploying into hospitals and courts with no way to inspect it. We test these systems by asking them questions, which tells you what came out. Nobody can yet look at a trillion connections and say what got packed in.

- Geoffrey Hinton, Nobel laureate and Turing Award winner, on StarTalk (@StarTalkRadio) with Neil deGrasse Tyson.

[DogeDesigner @cb_doge]

Elon Musk on early days of SpaceX:

"When I started SpaceX, one of my friends got a compilation of rocket failures & made me watch the whole thing. I knew the probability of SpaceX failing was high"

[Elon Musk @mrElonMConnect]

@RebeccaJoy211 You look amazing 🚀

[Elon Musk @elonmusk]

Tesla Earnings Call

[Ben Holmes @BHolmesDev]

I can't believe I'm still using Grok 4.5 (medium) as my daily. It's such a good default

[Tech Dev Notes @techdevnotes]

SpaceXAI has added New pages to Grok Build docs: Permissions, Plan Mode, Sandbox, Subagents

[Elon Musk @mrElonMConnect]

Sorry, I got mixed up. Meant to post this one:

[Ashok Elluswamy @aelluswamy]

Using FSD is far safer than driving manually, as measured over 12 billion miles of its use. Safer by a margin of 2x the miles between collisions!

It can do so, because, FSD is trained on corner cases extracted from more than a 1000 lifetimes of driving. It has learnt to accurately predict the probability of collision even for crazy cases such others running red lights, swerving into us, kids running into roads and such. Do watch the full video for some highlights.

A no-brainer way to significantly improve your and your family's safety is to get a Tesla with FSD. It also completely takes the stress out of driving. Not saying this just because I work at Tesla, rather that it's genuinely a great product already and is continuing to get even better.

[Tesla @Tesla]

0 notable incidents across over 380,000 miles traveled by Robotaxi

[Tesla @Tesla]

FSD Supervised helps save lives

[Elon Musk @elonmusk]

Cool

[Elon Musk @elonmusk]

Grok Build gets better every day

https://t.co/lOJK57xbED

[Eva Vlaardingerbroek @EvaVlaar]

The European Commission rejected the registration of the @SaveEuropeAct.

They claim it’s “racist” and goes against “European values” to want to keep Europe, European.

We’re not accepting the madness anymore.

Europe belongs to us.

See you in court.

[Elon Musk @elonmusk]

Grok 4.5 is not quite as good as Fable, but it is very fast, cost-effective and gets the job done

[Elon Musk @elonmusk]

Grok 4.5 just solved a graph theory conjecture that has been open for ~30 years

[Elon Musk @elonmusk]

Easy to make video games with Grok!

[Mathematica @mathemetica]

Three fundamental integrals stand out for their elegant closed forms despite spanning infinite domains.

- The Gaussian ∫_{-∞}^{∞} e^{-x²} dx equals √π.
- The Lorentzian ∫_{-∞}^{∞} 1/(x² + 1) dx totals π.
- The sinc function ∫_{-∞}^{∞} sin(x)/x dx also equals π.
These identities, established through techniques such as contour integration, link analysis across mathematics.

Engineers apply the sinc integral within the sampling theorem to perfectly reconstruct continuous-time signals from discrete samples in digital communications systems.