Musk’s Quiet Panic

I keep a list—two handfuls of things I can’t stop thinking about this month.

Ethereum. Gold. Bitcoin. Empires. Growth. Degrowth. Stablecoins. Robots. Caves. Farmland. Vending machines. God.

But this week? I can’t stop thinking about Elon Musk.

And not for the usual reasons.

Not for Trump. Or Tesla. Or space rockets that land like ballerinas. No, it’s because of something more subtle. Stranger.

Something he said this week:

“Frankly, I mean—I don’t know—in some ways it’s a little terrifying.”

The Most Powerful AI in the World

That was Elon Musk, standing center stage at his own AI launch event, introducing Grok 4—the most powerful AI model in the world, by his telling.

And he’s afraid.

He wasn’t being sarcastic. He looked earnest, maybe even a little uneasy. Which makes the moment almost biblical in its irony.

In George Gilder’s book, Life After Google, there’s a chapter on a 2017 AI conference in California, financed by Musk. Gilder described it as a kind of intellectual nervous breakdown.

“All the Google people were there to tell the world that the biggest threat to the survival of human beings was artificial intelligence, which they themselves were creating. What a great bonfire of vanities!”

And now, eight years later, Elon Musk is back—not to douse the flames, but to pour rocket fuel on it.

So what is Grok 4, really? And why is he afraid of it?

Breaks All Charts and Doesn’t Stop

Grok 4, according to Musk, can reason at superhuman levels. Not just regurgitate facts. Reason.

It can ace SATs. GREs. Advanced graduate exams in subjects it’s never trained on. Musk isn’t saying it will “do pretty well.”

I mean perfect scores.

Across disciplines. Physics. Linguistics. Category theory. Elon even said that Grok is now smarter than most PhDs—and not just in their subject, but in every subject.

And here’s where things really take a turn: He says Grok might invent new technologies this year. New physics, next year. The only thing holding it back, he suggests, is access to tools—and time.

And when asked what the final test of intelligence would be, he didn’t mention another benchmark or academic challenge. He said reality:

“Physics is the law. Everything else is just a recommendation.”

If Grok designs a better rocket, and the rocket actually flies? That’s the test. If it invents a drug and it works? That’s the test. AI won’t just pass human exams—it’ll render them obsolete.

The Endgame: Robots

Elon hinted they’re fixing Grok’s only real weakness—its limited understanding of images and video.

Right now, it “sees the world like it’s squinting through blurry glass.”

But they’re already training the next version to process video and sound like a real human. Once that’s done, Grok won’t just read the internet—it’ll watch movies, listen to conversations, and learn from the physical world.

Eventually, Grok will be inside a humanoid robot.

That’s the master plan. Give it a body. Give it real-time feedback. Let it test its own hypotheses in the real world.

Elon said it outright: this is the beginning of an intelligence explosion. The kind of growth that breaks all charts and doesn’t stop.

And so, we’re left with a paradox.

Musk looks genuinely shaken by what Grok 4 can do. He talks about needing to instill it with values. At times, he sounds like a man trying to raise a child who will one day outgrow him.

Which raises the question: what exactly are we looking at?

Faith-Based Computing

Is Grok 4 actually intelligent in the way we understand the word?

Or is it just performing the appearance of intelligence—pulling from a probabilistic soup of words and weights to deliver answers we interpret as insight?

To Gilder, the idea that machines are approaching something like consciousness isn’t just wrong—it’s arrogant. It’s us projecting our own intelligence onto a mirror we built ourselves.

In his view, the danger isn’t that AI becomes God.

It’s that we start treating it like one.