Empathetic AI

At Normally, we’ve been exploring the potential impact and implications of AI having an increased ability to understand and interpret people’s emotions, moods, and facial expressions.

Empathetic AI has vast potential to reshape industries and impact how we live, work, communicate, and who or what we trust.

Consider therapeutic AI chatbots in healthcare, or AI tools that recognise dips in mood to encourage healthier engagement with social media. Even finance, where emotional AI could read business leaders’ expressions for confidence, insecurity, or humility.

Already, we’re seeing startups like Hume AI mapping people’s emotions in new ways, using ‘semantic space theory’. This is different to, and arguably more complex than, the six or seven core emotions we traditionally use to categorise people’s behaviours and reactions.

The emergence of empathetic AI also raises questions about what it means to have empathy. Is it a uniquely human trait and ability? And what of the nuance that exists between countries and communities?

For example, facial expressions can mean different things depending on your nationality or culture. According to a Harvard Business Review study, a smile can mean one thing in Germany and another in Japan. The same HBR study found that emotional AI can assign more negative emotions to people of certain ethnicities than others. With any form of empathetic or emotion-reading AI, there’s the risk of bias.

And then there’s language, with more than 7,000 spoken globally, each with unique traits and cues. Language plays a huge role in how we understand our emotions, and depending on which language you speak, there are more — or fewer — words to describe various emotions, to varying degrees of specificity.

Accounting for the complexities of human emotions, language, and facial expressions, we must ensure similar representation, diversity, and data reliability when training and developing empathetic AI.

By taking such a mindful approach, tomorrow’s AI assistants could help us to form greater understanding, connection, and trust — not only between each other, but also between people and technologies.

What are the design or experiential opportunities ahead as we look to a future in which AI can intuitively understand human emotions?