Multi-modal artificial intelligence (AI) in personality forecasting can have revolutionary effects in hiring,
psychological diagnosis, and anthropomorphic interaction. Deep learning architectures based on audio, visual, and
textual data via behavioral cues tend to be efficient when the correct hypothesis is stated in the context of the Five
Factor Model (OCEAN)or by deciphering benchmarks (e.g., the approach of the ChaLearn First Impressions V2
data). Nevertheless, such systems will threaten to heighten demographic biases (e.g., gender, ethnicity) due to the
lack of representation, subjective labeling, or incorrect spurious correlations of the AI.