The Missing Step Between AI Hype and Actual Profit

Back in February, I picked up a flyer at an anti-AI march in London. It read: “Step 1: Grow a digital super mind. Step 2: ? Step 3: ?”

The group behind it, Pause AI, had unintentionally channeled the underpants gnomes from South Park—you know, the ones whose business plan was “Phase 1: Collect underpants. Phase 2: ? Phase 3: Profit.”

That meme has been dead for years, but it perfectly describes where AI is right now. Companies have built the tech (Step 1). They promise transformation (Step 3). Nobody agrees on what happens in between.

Pause AI thinks Step 2 should be regulation. Fair enough, but they’re vague about what that actually looks like. The AI boosters, meanwhile, skip straight to Step 3 being salvation. OpenAI’s chief scientist Jakub Pachocki told me a few weeks ago we’re racing toward an “economically transformative technology.” He knows where he wants to go—sort of. It’s hazy and still far off. Everyone’s taking a different route.

The problem is that for every grand claim, there’s a sobering reality check. Take two recent studies. Anthropic published a paper predicting which jobs LLMs will affect most. Their take: managers, architects, and media people should worry; groundskeepers and hospitality workers, not so much. But those predictions are basically guesses based on what LLMs seem good at in controlled tests, not how they actually perform in messy workplaces.

Then there’s the study from Mercor, an AI hiring startup, that tested agents from OpenAI, Anthropic, and Google DeepMind on 480 real workplace tasks done by bankers, consultants, and lawyers. Every single agent failed most of its duties.

Why the disconnect? Start with who’s making the claims. Anthropic has skin in the game. Most of the people telling us something big is coming are basing that on how fast AI coding tools are improving. But not every job involves coding. Other research shows LLMs are terrible at strategic judgment calls.

And when you actually deploy these tools, they don’t land in a cleanroom. They have to work alongside people and existing workflows. Sometimes adding AI makes things worse. Sure, maybe those workflows need to be torn up and rebuilt around the new tech, but that takes time and guts—neither of which the industry has in abundance.

That big hole where Step 2 should be? It’s an information vacuum. Nobody agrees on what’s about to happen or how. So the vacuum gets filled by whatever wild claim goes viral this week. A single social media post can shake markets. Evidence is optional.

We need fewer guesses and more evidence. That means transparency from model makers, coordination between researchers and businesses, and new ways to evaluate what really happens when AI rolls out in the real world.

The tech industry—and the global economy with it—is betting that AI will be transformative. That’s not a sure bet yet. Next time you hear bold claims about the future, remember: most businesses are still figuring out what to do with their underpants.