From cultures of experimentation, to culturally inclusive experiments

Thanks to the power of AI, computing experiences aren’t just becoming smarter - they’re becoming more human-like. Academic and industry research shows that these anthropomorphic AIs elicit emotional connections and expectations from their human users. We humans lean into these connections; we put googly eyes on our Roombas, eyelashes on our car headlights, and even marry our robots. 

‍

Fostering appropriate and ethical emotional connections with AI requires companies to build cultures of careful experimentation. It’s not just as simple as running A/B tests, for three reasons: 1) Tech companies have paid less attention to users’ emotional needs than their cognitive and functional ones, and therefore have less expertise in meeting emotional needs; 2) Users’ emotions vary widely across regional, ethnic, gender, socioeconomic, and other cultures, and experiments are needed to tune AI products to particular contexts, and 3) It’s easier to get emotions wrong than to get them right. 

‍

In this talk, we’ll describe and discuss these issues, talk about how experimentation needs to change in this brave new world of AI-powered experiences, and propose a series of experiments that can guide the development of emotionally appropriate, culturally inclusive AI.

‍