Willem van der Maden

In an era where artificial intelligence (AI) permeates every facet of our lives, this article presents and evaluates the positive AI design method aimed at supporting wellbeing in AI development.

This study benchmarks the emotional alignment of generative AI models by assessing their ability to express human emotions through imagery. Findings highlight variations in alignment by model, emotion type, and context, informing the design of emotionally expressive AI systems for wellbeing and reducing the risks of emotionally manipulative AI.