Ethical Affective Computing in the Workplace

Abstract

Affective computing — the detection and interpretation of human emotional states from behavioral and physiological signals — presents profound opportunities and risks in workplace contexts. When applied through surveillance mechanisms, it violates privacy, creates fear, and generates biased outputs. When applied through opt-in, privacy-preserving, aggregated approaches tied to development rather than monitoring, it can meaningfully improve team wellbeing. This paper defines the ethical framework that distinguishes responsible affective computing from workplace emotion surveillance.

Key Findings

• Affective signals collected through surveillance mechanisms (webcam analysis, keystroke patterns, email sentiment) produce biased, inaccurate, and harmful outputs.

• Opt-in self-report combined with behavioral signals produces more accurate emotional context data than passive surveillance.

• Aggregation is essential: individual-level emotional data should not be visible to managers without explicit employee consent and purpose justification.

• Transparency about signal collection, use, and data governance is a prerequisite for ethical affective systems.

• Development-oriented framing — using emotional signals to improve coaching, not to evaluate performance — is required for ethical and effective deployment.

How this connects to RUDY