Ethics Sheet for Automatic Emotion Recognition and Sentiment Analysis

Emotion recognition systems have dual-use potential: While these technologies can improve public health and commerce, they can also be weaponized to suppress dissidents and manipulate voters, requiring developers to carefully consider potential harms before building such systems.

Hidden assumptions shape emotion recognition systems: The paper identifies fifty ethical considerations that expose often-overlooked assumptions embedded in how emotion recognition is framed, including choices about data collection, methodology, and success metrics that can introduce bias and unfairness.

Privacy and social equity are major concerns: Automatic emotion recognition raises significant privacy risks and can disproportionately impact certain social groups, requiring special attention to how these systems might discriminate against or harm vulnerable populations.

"Why automate?" should come before "how to automate": The paper emphasizes that developers should engage in ethical reflection about whether automation is necessary and beneficial before making technical choices, rather than treating ethical considerations as an afterthought.

The affective computing community must actively engage with ethics: Rather than leaving ethical concerns to external oversight, researchers and developers building emotion recognition systems have an imperative responsibility to proactively think through the ethical ramifications of their work.