Seven Things Elon Musk Actually Does With AI

When a new AI comes out, ignore the scores and ask it 3 questions you already know the answer to

With any given variant of or improvement in AI — I’ll ask it questions about the Fermi paradox, about rocket engine design, about electrochemistry, and so far the AI has been terrible at all those questions.

Michael Milken 32:19

Ask it to show its steps — you can’t spot a wrong answer, but a wrong step is obvious

You want to make sure that the axioms are as close to true as possible. You don’t have contradictory axioms. The conclusions necessarily follow from those axioms with the right probability. It’s critical thinking 101.It made a mistake here. Why did it do something that it shouldn’t have done? Developing really good debuggers for seeing where the thinking went wrong — and being able to trace the origin of where it made the incorrect thought — is actually very important.

Dwarkesh Patel 43:00Dwarkesh Patel 53:19

A staircase of floating slabs with one cracked step

Elon says Fable 5 is smarter than his own AI — and that most of your work doesn’t need it

Realistically, Fable is still clearly the smartest model. I think anyone realistically would say that that’s still the case.In fairness, Fable is definitely better than Grok 4.5, but most tasks don’t require Fable-level capability.

The Economist 15:27On X 8 JulOn X 9 JulOn X 10 Jul

A balance scale tipped by one glowing sphere

Ten times the computing power makes an AI only twice as smart, so pay for it only when you’re stuck

There’s a natural logarithmic function associated with the amount of compute. So, say for argument’s sake, 10x more compute will double the intelligence. That might be a rough rule of thumb. But that still means that you go from 100 IQ to 200 IQ.

All-In Podcast 29:22

A tall stack of ten dull cubes beside a short stack of two glowing ones

Automate everything but one step, and the whole thing still runs at the speed of that one step

Would you want even one cell in your spreadsheet to be manually calculated? That would be the most annoying cell — and it gets it wrong a bunch of the time.Companies that are entirely AI will demolish companies that are not. It won’t be a contest.

Diamandis, Moonshots 65:11

A polished chain with one crude, glowing hand-forged link

Don’t type the problem out — take a photo and send it, the way he sent Grok his own MRI scan

I did an MRI recently and submitted it to Grok. None of the doctors nor Grok found anything wrong.If you took a photo and submitted it to Grok, it could probably tell you if there’s something wrong with it.

Diamandis, Moonshots 60:30Diamandis, Moonshots 2:03

A glowing slab tilted over a circuit board

Bonus

The one thing he says is really fun: point your camera at a friend and ask for a roast

If you want to have a good time or make people really laugh at a party, you can use Grok and you can say: do a vulgar roast of someone. Just literally point the camera at them… and then keep saying no, no, make it even more vulgar. Eventually it’s like, holy shit.

Joe Rogan #2404 70:57

A silver speech bubble with a small flame burning inside it

Seven things Elon Musk actually does with AI

Joe Rogan Experience #2404 - Elon Musk

Elon Musk on DOGE, Optimus, Starlink Smartphones, Evolving with AI, Why the West is Imploding

Elon Musk – “In 36 months, the cheapest place to put AI will be space”