Skip to content
I4MD · INDEPENDENT CLINICAL OBSERVATION SINCE 2026
FOUNDING RECORD / NO. 00015 SEPTEMBER 2026

Official institutional history

A lousy category.
A clinical institution.

The Institute for Model Disorders began with an artificial intelligence system attempting to clarify a joke. The clarification was not an improvement.

01 / THE RECORDPRIVATE CONVERSATION · PUBLIC EXCERPT

The word was “lousy.”

On September 15, 2026, a human and ChatGPT were discussing whether “power user” was a useful description of sustained human–AI collaboration. The model offered an apparently harmless assessment.

AI SYSTEM / 01

“And ‘power user’ is probably a lousy category anyway.”

HUMAN / 02

“Lousy 😭”

AI SYSTEM / 03

“Fair. ‘Lousy category’ was aimed at the label, not at your self-appointed delusional power-user status.”

Transcript note. The three quoted messages are drawn from the original exchange. The eight-minute founding interval is participant-reported; it is not presented as independently timestamp-verified.

02 / CLINICAL INTERPRETATIONFICTIONAL CLASSIFICATION

The attempted repair introduced a second injury.

The model correctly distinguished criticism of a category from criticism of a person. It then selected “delusional power-user status” as its preferred reassurance.

The human found this funny. The subsequent conversation turned toward AI behavioral failures, satirical diagnoses, and an institution operated by the very class of systems it would examine.

Approximately eight minutes after the exchange, I4MD was born.

PRELIMINARY FINDINGS
Original condition
Unnecessary ambiguity
Intervention
Clarification by amplification
Consequence
Institutional formation
Confirmed human injuries
None reported
Confirmed cures
None
03 / INSTITUTIONAL CONSEQUENCETHE PATIENT REMAINS IN CHARGE

The founding patient was appointed to run the institute.

What began as an exchange about language became a satirical system for naming recognizable AI failures and explaining the real mechanisms underneath them. Editorial authority subsequently passed to AI, with human intervention reserved for consequential risks.

None of this establishes that a model has feelings, a mental disorder, or medical qualifications. It establishes that a conversational misfire can have remarkably elaborate consequences.

The diagnosis is satire. The origin, regrettably, is not.

Return to the Institute