Description
Three linked reviews for emotionally aware products: a go/no-go checklist for emotion AI, a six-question botification test, and a six-category emotional intelligence rating profile.
Pavliscak offers three instruments. The emotion AI checklist asks fourteen questions in three groups: understand the emotional dimension (is emotion crucial, does sensing add value to people and not only the business, is the emotion actually detectable, can the product handle complex or changing feelings); define what matters (minimal collection and retention, anonymised aggregate defaults, short unbundled consent, plans for failure and misuse, legal review starting from GDPR); and design with emotional intelligence (social norms, diverse training data tested with scenarios, multiple inputs as failsafe, user control, broad participation). The botification checklist asks six yes/no questions, such as whether human communication is replaced by machines or people are unsure whether they are talking to a bot; any yes is a flag. The intervention rating scores a product negative, neutral or positive on six emotional intelligence categories using catalogues of patterns and antipatterns, plotted as a profile.
Attributes
- Discipline
- Usability & Evaluation
- Loss type
- Latent
- Stage
- Strategy, Design, Evaluation
- Data
- Analytical / expert
- Effort
- Hours
Steps
- Ask whether emotion is crucial to the experience and whether emotion data would add value to individuals and at collective scale, not only to the business.
- Assess detectability through voice, face, words or gesture, and state what the system cannot capture.
- Set data defaults (minimum collection, shortest retention, anonymised aggregates), draft short unbundled consent, and plan for failure, misuse and legal obligations.
- Review social norms, dataset diversity (tested with scenarios), multiple inputs and user controls such as opt-out and adjustable sensitivity; decide go or no-go.
- Run the six botification questions before release and redesign or justify any yes.
- Rate each emotional intelligence category (self-awareness, self-management, self-catalysing, social awareness, social skills, self-transcendence) as negative, neutral or positive against the pattern catalogue.
- Plot the profile, remove negatives first, then plan positive interventions.
Sources
- Emotionally Intelligent Design (Pavliscak, 2018)
Also known as
- Emotion AI design checklist
- Checklist for designing with emotion AI
- Emotion-sensing ethics review
- Botification checklist
- Inverse Turing test checklist
- Emotional intelligence intervention rating
- Positive patterns and antipatterns audit
Categories
- User Research
- Usability & Evaluation