AI agents are not employees

Your AI risk register is being written by a metaphor.

Many companies assume, or directly call, AI agents a "digital workforce" in presentation decks when suggesting replacing humans with AI agents. The board believes it, HR frameworks are being stretched over it, and every assumption in that metaphor is wrong in a way that costs money.

When you call an agent an employee, you silently import two centuries of assumptions. Line them up against the actual participant:

• Employees tire, which rate-limits them. Agents don't, so nothing does.

• Employees fear consequences, which makes them cautious. Agents don't, so caution must be engineered.

• Employees learn from yesterday's mistake by this morning. Agents learn nothing between deployments unless you built the loop.

• Employees err independently: two analysts are rarely wrong the same way at the same moment. Agents err in sync, with their own past outputs and with every other agent on the same model.

And when an employee acts, a legal person owns the action. When an agent acts, accountability is answered by an architecture, or not at all.

Each broken assumption is a risk the employee frame can't see, and behind them all sits the strangest property of the new participant: agents are the first non-human actors to generate probabilistic decisions with consequences.

Every computer before them accelerated human decisions, and errors meant malfunction. An agent's errors are the output of normal operation.

No fatigue means those errors amplify at machine speed and nothing absorbs them.

Probabilistic actors at scale added variance in your operations: your planning horizon quietly shortens.

Synchronised errors mean your firm fails to keep pace with your competitors, long before any HR-style "performance review" would notice.

Agents are not better employees or worse employees. They're a different kind of economic participant, and every management instrument you own was built for the other kind. HR metaphors don't simplify the picture for the board. They point the board at the wrong toolkit.

The full argument is out in the article "Agents Are Not Employees", on the three systemic risks the workforce metaphor hides: amplification, entropy, coupling.

The full argument, on the three systemic risks the workforce metaphor hides: amplification, entropy, coupling, is Chapter 11 of Architecture of Intellect.

What's the most dangerous "digital workforce" assumption you've heard presented to a board with a straight face?