Disrupted Emotion – how Artificial Intelligence is being taught anger

Early stage anger simulation intended to teach support workers how to deal with angry customers – but how will it end?

iDisrupted Commentary

Data scientists from New Zealand are teaching an AI to learn anger. Sounds crazy, but the idea is to help companies deal with common customer complaints.



Tech firm, Touchpoint Group is using machine learning to help its AI system recognise – and even simulate – anger. Called Radiant (a name taken from Isaac Asimov’s Foundation series), the purpose of the $400,000 project is to develop an automated system that can defuse angry customer service calls. Once complete, the system will be capable of generating over a hundred million (!) angry interactions. Touchpoint’s data scientists are hoping to build a system that can autonomously find the best response to typical customer complaints.

But as The Telegraph’s Olivia Goldhill explains, Radiant won’t be capable of actually experiencing genuine anger. It can only be programmed to mimic and repeat these nasty conversations. As Stuart Armonstrong from the Future of Humanity Institute told her:

There’s not much variety in human anger. If someone’s angry they’ll just hurl insults at you, there’s not much subtlety of interaction so you don’t have to code anything complicated. Anger is easy to imitate without having to go into depth.

Touchpoint, which expects to roll-out a live system by the end of the year, is targeting big banks, telcos, and insurance companies as potential customers.
So having taught computers to simulate anger where does it end? There is already a Google for Robots – maybe they can learn AI anger from them. And AI is learning emotion, and even robots are learning trust and regret