A study crossed the wire this week that puts a number on something most people feel and can’t prove.
Researchers at RWTH Aachen University in Germany surveyed 1,110 ordinary citizens and 119 AI experts. Same questions. Seventy-one scenarios about where automation takes us in the next ten years. Cars that drive themselves. Software that tutors children. Systems that make medical calls, political calls, life-and-death calls.
The two groups looked at the same future and saw two different worlds.
Here’s the number that matters. When the experts judged whether an AI application was good or bad, the expected benefits drove their judgment about three times as strongly as the risks did. Read that again. A specialist could see high risk in a system and it barely moved the needle — as long as the benefits looked good, the enthusiasm held.
The public ran the opposite math. Everyday people prize convenience too. But when they sensed danger in a system, that danger dragged hard on their approval. Risk counts, heavily, when you’re the one who has to live inside the result.
Two groups. Two formulas. And only one of them is building the machines.
The researchers gave the danger a name, and it’s worth learning: procrustean AI.
The name comes from the Greek myth of Procrustes, a blacksmith with an iron bed. He offered travelers a place to sleep. If the guest was too short for the bed, he stretched them. Too tall, he cut them down. The bed never changed. The guest always did.
Procrustean AI is what you get when engineers build systems from their own optimistic picture of the world — and everyone else has to stretch or shrink to fit. The system doesn’t bend to human needs. The humans bend to the system’s constraints.
If you’ve ever fought a phone tree that wouldn’t hear your problem, or an automated decision with no appeal, or a feed that decided what you’d see without asking — you’ve slept in the iron bed. You know the size of it.
Now look at what the two groups feared, because this is where the study stops being interesting and starts being important.
The public’s deepest concerns were the big ones. Loss of human control. Machines replacing human relationships. Systems controlling information. Software making independent decisions about human life.
The experts rated those same scenarios as unlikely and moved on. Their optimism pointed elsewhere — healthcare gains, scientific progress, environmental tools. Real benefits, honestly foreseen. But the fears of the people who will actually live with these systems were, in effect, filed under improbable.
Here’s the problem with that filing. The public’s fears aren’t a forecast of specific events. They’re a demand for a specific guarantee: that a human stays in control, that the machine answers to the person, not the other way around. You don’t answer that demand by calculating probabilities. You answer it by building the guarantee.
The study’s own conclusion says as much. The researchers call for participatory design — systems shaped by the people who have to use them, not just the specialists who ship them. If the future is built solely from the expert’s mental model, they warn, it will neglect the valid concerns of everyone else.
That’s the field, in its own journal, saying the missing ingredient is a governance framework written from the public’s side of the gap.
Read that call carefully, because here’s the plain truth of it: that framework exists. It’s been in the public record for over a year.
The Faust Baseline wasn’t written in a lab. It was written by a builder — decades of industrial and construction work, a lifetime of watching what happens when the people who design a thing never have to stand under it. It was written from the guest’s side of the table, holding exactly the priorities the study says the public holds and the experts discount.
Look at the match, point by point.
The public’s deepest fear is total loss of human control. The Baseline’s load-bearing principle is the human in the room. The operator holds ratification authority over everything. Nothing enters the permanent record without explicit human approval. The machine proposes; the human disposes. That’s not a feature. That’s the foundation — it’s called PMAP-1, and every one of the other twenty protocols stands on it.
The public fears systems that decide without accountability. The Baseline requires every constraint named, every boundary disclosed, every claim backed by evidence present in the session — and a standing challenge right, appended to every substantive response, so the human can test the machine’s output before accepting it.
The public fears the iron bed — being forced to fit the system. The Baseline is the opposite architecture by design: the framework travels with the user, not the platform. The user’s standards load first and supersede platform defaults. Full portability, no lock-in. The bed adjusts to the guest. That’s the rule.
The experts weigh benefits three-to-one over risks. The Baseline weighs them the way the public does — with protocols that flag irreversible recommendations before they’re delivered, that stop when evidence ends, that refuse to let a coherent story stand in for missing data.
Every fear the study measured, the framework answers. Not with reassurance. With structure.
And tomorrow — July 4, 2026 — that structure completes its current build. The Agentic Governance Protocol ratifies into the permanent Codex: the gate that governs what happens when AI systems act on their own, the exact scenario the public in this study feared most and the experts waved off.
The date is not an accident. A framework built on the principle that the user governs the machine, ratified on Independence Day, while the field’s own research confirms the governed have been left out of the room.
The experts are building the bed. They mean well — the study shows their optimism is genuine, their benefits real. But good intentions never changed the size of an iron bed.
Someone had to build for the guest.
Someone did. It’s on the record, dated, and the flag goes up in the morning.
Post Library – Intelligent People Assume Nothing
The Faust Baseline™ — intelligent-people.org
Codex 3.5 | Twenty Protocols | Ratified and dated on the public record.
Contact: micvicfaust@gmail.com
Purchasing Page – Intelligent People Assume Nothing
© 2026 The Faust Baseline LLC | All Rights Reserved
