Editorial illustration for Atlantic report: MorphCast AI tags employee emotions during boss meetings
Atlantic report: MorphCast AI tags employee emotions...
Your boss’s meeting room just got a new attendee: an algorithm that claims to read your soul from your face. Ellen Cushing volunteered to test MorphCast, an emotion AI tool that scanned her expressions during a one-on-one with her supervisor. The verdict?
She was amused, determined, interested, and, every so often, impatient. But this wasn’t a quirky stunt. It’s a window into a booming industry that now analyzes job interviews, call center chatter, and even the heartbeats of office chairs.
MetLife listens to its agents’ vocal pitches. Burger King is piloting a headset named Patty that rates friendliness. And Slack integrations quietly assess your messages for sentiment.
The problem? The science underpinning much of this surveillance is shaky at best. Paul Ekman’s classic six-emotion model has been widely debunked as simplistic.
Neuroscientist Lisa Feldman Barrett points out that facial movements don’t carry inherent emotional meaning, people scowl when angry only about a third of the time. Worse, studies show these tools disproportionately misread Black faces as angry. Pseudoscience, embedded in workplace tech, is already causing real harm.
The machines are watching, and they don’t know what they’re seeing. A smile can be a weapon, a mask, or a nervous tic; a scowl a moment of focus, not fury. Yet the algorithms charge ahead, turning flickers of the face into verdicts on character, performance, and potential.
The science is weak, the bias is real, and the surveillance is unchecked. Workers are being judged by a pseudoscience that can’t tell the difference between amusement and impatience, let alone between anger and a smile on a Black man’s face. This is not a tool.
It is a system that trades human complexity for a false sense of control. The harm is already here. The question is whether we will let it write the rules.
Related Reading
Policy & Regulation Grammarly faces class-action suit over AI ‘Expert Review’ feature
Policy & Regulation Claude Mythos highlights EU AI safety gaps, says researcher Caroli
Policy & Regulation After ditching AI fitness apps and a Fitbit, I return to Peloton classes
From Archives Musk lawsuit scrutinizes OpenAI safety as tech gains power under scrutiny
From Archives Trump renames US AI Safety Institute to Center for AI Standards, dropping “safety”
Common Questions Answered
What is MorphCast and how does it analyze employee emotions?
MorphCast is an emotion AI tool that scans facial expressions to detect and categorize emotional states during workplace interactions. The technology analyzes micro-expressions and facial movements to assign emotion tags such as amusement, determination, interest, and impatience, as demonstrated when journalist Ellen Cushing tested the tool during a meeting with her supervisor.
Beyond meetings, what other workplace applications is emotion AI being used for?
According to the article, emotion AI technology is expanding across multiple workplace contexts including job interviews, call center monitoring, and even analyzing heartbeats from office chairs. This widespread adoption demonstrates how the emotion recognition industry is growing beyond traditional meeting room surveillance into various employee monitoring scenarios.
What are the main concerns raised about emotion AI accuracy and bias?
The article highlights that the science behind emotion AI is weak and the technology exhibits real bias issues that remain largely unchecked. The algorithms struggle to distinguish between different emotional states—for example, confusing amusement with impatience or failing to differentiate between anger and a smile—yet they are still being used to make judgments about worker character, performance, and potential.
Why is the article skeptical about using facial expressions to judge worker performance?
The article argues that facial expressions are inherently ambiguous and context-dependent, noting that a smile can be a weapon, a mask, or a nervous tic, while a scowl might indicate focus rather than fury. Because algorithms cannot reliably interpret these nuanced human expressions, using them as the basis for performance evaluations represents a form of pseudoscience that puts workers at risk of unfair judgment.
