I have no shortage of worries, but AI taking over the world doesn't really rank. Maybe it's just me with my head in the sand. But two reasons say otherwise: the predictions are usually wrong, and I prefer to focus on what I can control.

Even if your work doesn't revolve around GPTs and agents like mine does, the recent panic about AI going rogue is hard to miss.

Or rather, panics. There's the Hugging Face incident (which was actually caused by rogue OpenAI agents). There's the Anthropic researcher who quit, saying that AI is "gambling with our lives." There are the CEOs of all major AI frontier labs saying "we must slow down" (even Elon agreed with Anthropic CEO Dario Amodei's call to action).

Panic-inducing on the surface. But here's how I see it:

The predictions are usually wrong

This fall's panic is rogue AI. But look at the year: the panic-of-the-month club has been flourishing, yet the panics keep contradicting each other.

In January, AI was taking our jobs. By March, the panic had flipped: we weren't using AI enough, and companies ranked employees by how many tokens they burned (tokenmaxxing). By May, we were exhausted from creating agents to "clear the endless backlog." By June, the panic flipped again: we were using AI too much. Uber engineers had burned the whole year's AI budget by April, and the labs were turning up the cost per token (now we're efficiency-maxxing). Here we are in September, with job creation up and AI changing jobs but not gobbling them.

But, don't take my word for it. Someone (who? tell me if you know) has been cataloguing the fear-mongering on AI in an index aptly named "boys who cried AI."

World-destroying rogue AI is of course a different level of angst than AI stealing jobs. But on prediction accuracy? I'd gamble they're about the same.

Part of why predictions miss: We hear them and change course. AI is taking jobs? We learn new ways of creating value. We fear AI is "gambling with our lives"? We get way more serious about guardrails.

We've been here before. As my friend Greg says, people were losing limbs in machines before OSHA set workplace safety guidelines.

Lest you think I've been watching all this from a beach chair: in January, as company after company announced layoffs, I worried about never working again. And I still don't completely trust that I've set up my agents not to turn on me.

Maybe rogue AI will take over the world, but it feels more likely it'll go the way of the job losses... it will shift the world, for sure. And we'll adjust, it'll adjust. World will go on.

But even if I'm wrong...

I prefer to focus on what I can control

I've had my phone by my side, but so far neither Dario, Elon, Sam, nor any world leaders have called to ask for my help in containing AI. No one wants my opinions on how to guardrail it to protect humanity. No one is looking at whether I use fewer tokens in protest.

So I'm continuing to push my own knowledge around AI. Building new things in new ways, learning where to trust it and where my wisdom matters. Helping my clients gain their own expertise in it.

While the panic-waves keep rolling in overhead, this is diving below the surface. That's where it's calm and we can get curious.

Mostly, I live down there.

So next time an AI panic crosses our feeds, we can ask: Is there anything here I actually control? If your answer is yes, give Dario my best when you speak! If not, we can dive under the wave and go build something.

See you below the surface.