Adding emotional intelligence to AI

Emotibot combines deep-learning AI, image and text-based recognition software to offer an open-source API for more emotionally aware chatbots.

The use of AI chatbots is on the rise, whether it’s helping users with recruitment, insomnia or acting as a hotel concierge, and now China-based Emotibot are enabling developers to offer more personable automated user experiences.

In collaboration with MIT Media Lab, Emotibot has been applying image recognition and a deep-learning approach to develop more emotionally intelligent chatbots. The API is able to pick up on 22 emotional patterns used in text conversation, 7 emotions from voice recognition and also 7 unique emotional states from facial expressions, so that if users have cameras and microphones activated on their personal devices, Emotibot will adapt it’s responses and interactions based on how the user feels in real-time. A feature called Emoti-eye will also enable the chatbot to recognize distinct types of clothing or objects also present through device cameras, which the API is capable of retaining a ‘memory’ of for subsequent interactions, with the chatbot constantly adapting via self-learning algorithms. The Emotibot API SaaS will be open source, allowing developers to fully customize user experience whatever the context.

Could emotionally intelligent chatbots be a feature in staff-less stores such as Amazon Go?