Abstract:This paper presents a new measure of emotional perceptiveness called PAGE: Perceiving AI Generated Emotions. The test includes a broad range of emotions, expressed by ethnically diverse faces, spanning a wide range of ages. We created stimuli with Generative AI, demonstrating the potential to build customizable assessments of emotional intelligence at relatively low cost. Study 1 describes the validation of the image set and test construction. Study 2 reports the psychometric properties of the test. Despite its brevity - 8 minutes on average - PAGE has strong convergent validity and moderately higher internal consistency than comparable measures. Study 3 explores predictive validity using a lab experiment in which we causally identify the contributions managers make to teams. PAGE scores strongly predict managers causal contributions to group success, a finding which is robust to controlling for personality and demographic characteristics. We also discussed the potential of Generative AI to automate development of non-cognitive skill assessments.
What problem does this paper attempt to address?
The problems that this paper attempts to solve mainly focus on the limitations of existing emotion - perception measurement tools. Specifically, the author points out that the current emotion - perception tests have the following three main problems:
1. **Ethnic Homogeneity**: Most tests use ethnically homogeneous stimulus materials, especially white faces, which may lead to test bias because participants are quicker and more accurate in recognizing emotions expressed by individuals with the same cultural background.
2. **Limited Emotional Range**: Many assessment tools only include a few basic emotions, such as anger, disgust, fear, happiness, sadness, and surprise, while ignoring complex emotions in real life.
3. **Practical Application Limitations**: Existing test tools are often long and costly, and are not convenient for use in field studies, especially in scenarios where cross - platform use is required.
To solve these problems, the author has developed a new emotion - perception measurement tool - **PAGE (Perceiving AI Generated Emotions)**. PAGE utilizes generative artificial intelligence technology (Generative AI), especially OpenAI's DALL - E 2, to generate diverse face images that express a wide range of emotions, including basic and complex emotions. The features of PAGE include:
- **Ethnic Diversity**: The images cover faces of multiple ethnic backgrounds, with an age range from 20 to 60 years old.
- **Emotional Richness**: The test includes 25 emotions, not limited to basic emotions.
- **Low Cost and High Efficiency**: Through generative AI, researchers can quickly generate customized test materials at a lower cost.
### Research Design
The paper verifies the validity and practicality of PAGE through three studies:
1. **Study 1**: Describes the process of generating the image set and test construction, including using DALL - E 2 to generate images, emotion selection, prompt engineering, standardization processing, and the verification process by human raters.
2. **Study 2**: Evaluates the psychometric properties of PAGE, including internal consistency and convergent validity. The results show that PAGE has a high internal consistency (Cronbach’s alpha = 0.73) and is highly correlated with the widely used RMET (Reading the Mind in the Eyes Test) (raw correlation = 0.66, disattenuated correlation = 0.88).
3. **Study 3**: Explores the predictive validity of PAGE. Through laboratory experiments, the study found that PAGE scores can significantly predict managers' causal contributions to team success, and this association remains robust after controlling for personality and demographic characteristics.
### Conclusion
Overall, PAGE, as a new emotion - perception measurement tool, overcomes the limitations of existing tools and provides a reliable, short, and emotion - perception assessment method suitable for diverse populations. In addition, this study also demonstrates the potential of generative AI in creating customized emotional intelligence tests. Future research can further improve AI technology to make test creation more efficient and convenient.