Artificial General Intelligence  (AGI)  has kind of taken over the AI conversation lately. Everywhere you look, someone’s talking about it.

It’s not just about science anymore. There’s money involved. A lot of it.

Tech companies are spending like crazy, energy costs are rising, and investors are expecting big results. So having a bold claim like “AGI is near” (or already here) sounds pretty convenient.

But here’s the catch. Everything depends on how you define AGI.

So… What Even Counts as AGI?

That’s where things get a little slippery.

Take Jensen Huang, CEO of NVIDIA  a company sitting at around $4 trillion valuation, powered largely by AI chips. In a recent conversation with Lex Fridman , he basically said… we’ve already achieved AGI.

Yeah. Already.

But wait. It’s not that simple.Because earlier, Huang had a different definition. Back in 2023, he said AGI would mean software that can pass tests similar to human intelligence at a competitive level. And he thought that might happen within five years.

Now suddenly… we’re already there?

A Convenient Definition Shift

Lex Fridman gave Huang a pretty interesting benchmark. He asked:

What if AGI means an AI that can build and run a billion-dollar company?

Sounds fair, right? Not too easy, not impossible either.

Huang’s answer?
“We’ve already achieved it.”

But here’s where it gets… a bit tricky.Huang reinterpreted the question. According to him, the AI doesn’t need to build a sustainable company. It doesn’t need to manage teams, deal with investors, or survive long-term.It just needs to hit a billion dollars once.

That’s it.

The Viral App Scenario

To explain his point, Huang gave an example. Imagine an AI creates a simple app. Nothing fancy. It goes viral. Millions  maybe billions of users jump on it. Each pays a small amount. Boom  it hits a billion dollars.

And then?
It fades away.

Huang compared this to the dot-com era, where many websites blew up quickly and disappeared just as fast.And honestly… yeah, that kind of thing could happen.

But is that AGI?

That’s the real question.

The Big Gap No One Can Ignore

Here’s something interesting and kind of important.

Huang himself admitted that even if you had thousands of AI agents, none of them could build something like NVIDIA.

Zero percent chance, he said.

That’s not a small detail. That’s everything. Because building a real company  a lasting one  isn’t just about launching something viral. It’s about decision-making, leadership, adapting over time, handling pressure, managing people… the whole messy human side of things.

And that? AI isn’t really doing yet.

So Are We Actually Close?

What Huang describes  a short-lived, viral AI product — is impressive. No doubt. But it’s not the same as the kind of AGI people imagine. You know, the one that can think, plan, create, and operate at a human level across everything.

That version? Still feels far away.

It’s like calling a spark a wildfire. Technically related… but not quite the same thing.

The Real Takeaway

AGI isn’t just a technical milestone. It’s also becoming a narrative. And depending on how you define it, you can say we’re close… or not even remotely there. Right now, it feels like the definition itself is doing most of the heavy lifting.

So yeah  maybe we’ve reached a version of AGI.

But the one that changes everything?
That’s still somewhere down the road.