The perfect corporate psychopath

In Descartes’ Error, Antonio Damasio describes patients with damage to emotional processing centers. Their cognitive abilities remained intact—they could analyze, calculate, reason. Yet they became ineffective, unable to complete tasks despite unimpaired intelligence.

The revelation: emotions aren’t obstacles to rational decision-making but essential components. Shame motivates correction. Pride drives quality. Fear prevents catastrophic mistakes.

Language models have no emotional substrate. They perform pattern matching that approximates expertise, but without the feedback loops that create accountability. No shame when wrong. No pride in accuracy. No fear of consequences.

They’ll produce incorrect medical advice with the same confidence as accurate information. Cite non-existent legal precedents. Write code with subtle bugs that surface months later. Always authoritative, never uncertain.

The perfect corporate psychopath.

When a developer feels unease looking at fragile code, that’s not weakness—it’s evolved pattern recognition. When an accountant double-checks figures because something “feels off,” that’s a form of intelligence no model can replicate.

Humans who fail repeatedly feel those failures. They improve or they leave. They can be held responsible, legally and socially. The emotional weight creates accountability.

An AI model feels nothing when it fails. It doesn’t learn from shame or fear consequences. The companies that produce these models have carefully constructed terms of service that disclaim responsibility. The burden falls on whoever trusted the output.

The absence of emotional response isn’t a feature—it’s a fundamental limitation. The emotions we call “irrational” are what prevent us from confidently making catastrophic errors. A system that can’t feel bad about failure can’t truly be trusted with critical decisions.