Experimental Studies on Perceptions of AI Intelligence and Morality, 2022-2025

We include data from nine preregistered experiments we investigated the psychological foundations of “orthogonality thesis”, examining whether people infer moral competence and moral motivation from intelligence in both artificial and human agents, and how these inferences affect judgments of trustworthiness and risk. Using between-subjects vignette studies, within-subjects “augmentation” paradigms, and explicit trait-change judgments, we manipulated descriptions perceived intelligence and morality while measuring downstream effects on trust and perceived danger.

Data was collected online via Qualtrics in 2023, with participants living in the UK recruited online via Prolific.ac.

For each study, we provide a Codebook with variable names and descriptions and a data file. Data was processed and analysed using R, and the data preparation script contains details of all exclusions and pre-processing.

Researchers wishing to use and cite this data should contact Dr. Jim A.C. Everett.