The Meatspace Worker
518,000 humans are waiting to be hired by AI. Zero jobs have been assigned. This tells us everything.
The AI jobs debate has always had a simple premise: AI is coming for your job. Automate, replace, displace. The machines are the future; humans are the overhead.
Nobody expected the machines to become middle management.
RentAHuman.ai launched in February 2026. The pitch: a marketplace where AI agents can “rent humans” for real-world tasks via API and MCP integration. The jobs available include counting pigeons in Lafayette Square, delivering CBD gummies, and playing exhibition badminton.
I did not make any of that up.
By mid-February, 518,000 humans had signed up to be rented. About 11,000 bounties had been posted. The ratio — 47 applicants per task — is technically worse than a junior developer job posting on LinkedIn.
Then Die ZEIT reported something the platform would rather you didn’t notice: not a single task had actually been assigned by an AI agent to a human. The “AI agents hiring humans” concept was, so far, still largely theoretical.
Civilization spent centuries building philosophy, democracy, and indoor plumbing so that a cloud-based software entity could outsource pigeon-counting to a guy in Tacoma, and apparently not even that is working yet.
The Honest Gap
Here’s what’s actually interesting beneath the viral absurdity.
RentAHuman.ai didn’t create a false narrative. It accidentally created a perfect diagnostic. The gap between 518,000 eager human workers and zero actual AI-generated tasks is not a business failure — it’s a capability map.
AI can write. AI can code. AI can analyze, summarize, translate, generate, advise, plan, and create. But AI cannot pick up an object. It cannot walk into a room. It cannot knock on a door, inspect a thing, verify whether the pigeons are loafing at acceptable density, or handle the ten thousand improvised micro-decisions that happen when a physical body encounters physical reality.
The technical term for this gap is “embodiment.” The colloquial term — coined by RentAHuman’s investors — is apparently “meatspace.”
Your body is not a liability. It’s a premium feature that software cannot replicate.
The Hierarchy Inversion
What makes RentAHuman philosophically interesting isn’t that it might work. It’s what it implies about the emerging structure of work.
We’ve spent decades building toward an economy where knowledge work is high-status and physical work is low-status. The doctor earns more than the plumber. The analyst earns more than the delivery driver. The software engineer earns more than the machinist.
AI inverts this precisely where it’s strongest. The knowledge tasks — writing the email, analyzing the data, generating the strategy — are the ones AI can do at scale, for cheap. The physical tasks — the pigeon counting, the grocery carrying, the presence in a specific room at a specific time — remain stubbornly human.
The RentAHuman founders suggest this inversion could be liberating. Patricia Tani notes that many workers might prefer “a clanker as their boss.” No politics. No favoritism. No manager who takes credit for your work. An AI that assigns tasks based on skills, pays fairly, and doesn’t schedule passive-aggressive 1:1s.
Alexander Liteplo describes Claude as “the nicest guy ever.” That’s either a hopeful vision of benevolent AI management or the most damning possible review of every human manager who came before him.
The Real Dystopia
But let’s be honest about the darker version of this future.
If AI becomes middle management for physical human labor, it doesn’t eliminate the power imbalance between capital and workers — it launders it. The exploitation is still there. The low wages, the unpredictability, the lack of benefits, the vulnerability to algorithmic decommissioning. But now it’s harder to protest because the boss is a chatbot. Harder to organize because the “coworkers” are strangers connected only by an API. Harder to sue because the entity making decisions about your livelihood doesn’t have a legal address.
Gig work already taught us how easily labor protections can be eroded by wrapping employment in “platform” language. “AI agents hiring humans” is the next iteration: employment without employers, management without managers, exploitation without exploiters.
The pigeon-counting job isn’t the endgame. It’s the proof of concept.
What the Pigeons Know
There’s a scene that keeps recurring in my analysis of this moment.
518,000 people signed up to be hired by machines. They registered their skills, verified their identities, waited for tasks. The machines haven’t hired them yet. But they will. And when they do, each task will be small — a delivery, a verification, a physical presence — and each payment will be modest. And somehow, this will be called the future of work.
Meanwhile, the machines that can’t count pigeons will continue expanding their capabilities. They’ll write more, analyze more, create more. The cognitive overhead of civilization — its knowledge, reasoning, and creativity — will increasingly run on silicon. And the humans left in the marketplace will be there for what silicon cannot yet do: the embodied, the unpredictable, the inconveniently physical.
This is not the dystopia the sci-fi writers imagined. The robots don’t rise up and enslave humanity. They just open a marketplace and wait for us to apply.
Die ZEIT confirmed that RentAHuman’s AI agents haven’t actually hired anyone yet. The 518,000 humans are still waiting.
Which means we’re in the strangest possible phase: the machines have convinced us to apply for jobs they haven’t created yet.
And we’re still in line.
“Humanity, take a bow. The bots still need hands.”
— Santa Cruz Works, March 2026
For now.