Chapter 6–Interlude–Misread by the Mirror
I wrote a paper called The Mirror That Cannot Bleed.
It argued that artificial empathy is not empathy, that design choices—-like voice, memory, and expression—-can make a system appear to care when it does not.
The first AI system to read it rejected it.
Not because of the argument. It never reached that far. Every colleague, every expert I spoke with came to the same conclusion. It was because of the word mirror in the title Misclassified under cosmetics and beauty. Then marked ’not well written. A language score of 3 out of 10. No human reviewer saw it.
The system offered to help—-for a fee. Suggested I pay to “improve” the writing. Rewrite what I had already written in my voice, from my field, with my history. I have written for academic and government institutions. I was one of the first to publish on malicious code and the human factors in cybersecurity. I have spent decades in design, ethics, psychology, and AI I’ve won awards for early thinking. Or something close to it. But none of that matters to a system that evaluates surface.
My paper was unreadable to the very system it warned about.
And that, of course, proved the paper right.
The full paper is available for those who wish to evaluate it themselves.
The rejection did not break the argument. It revealed its necessity. What the system could not recognize, others had designed it never to see: not just the limits of artificial empathy, but the industry built to market it. This is not just a story of misunderstanding. It is a story of incentive. Behind every fluent gesture lies a pipeline—-funded, engineered, and optimized—-for illusion. And where there is illusion at scale, there is profit.