Lana Novikova is on a mission to teach computers to “read” your emotions.
Why is this important? Her Toronto start-up, Heartbeat AI, is connecting the dots between artificial intelligence, cognitive sciences, consumer research and marketing to help organizations better know their customers, employees and patients – by understanding how they feel.
Heartbeat AI’s award-winning platform analyzes text – in near real-time – to extract emotional words and phrases which are then grouped to recognize the writer’s emotions. Text can be extracted from any source, including open-ended survey questions, call centre transcripts, customer feedback, product reviews and employee comments, and then turned into a user-friendly dashboard within minutes.
The software is designed to recognize 99 complex emotions, which are then clustered into nine primary feelings – Joy, Love, Trust, Anger, Fear, Disgust, Sadness, Surprise and Void (a lack of emotion). Novikova’s goal is to get at the “deep why” underpinning our decisions.
A market researcher by trade, Novikova realized that asking open-ended questions gave deeper feedback than the commonly used closed-ended questions. She says that while the information she received was invaluable, what she lacked was a way to analyze this big data.
“I realized what I needed and what the industry needed was a tool that will take all this text, all the words and stories, and separate them into meaningful chunks. And for me, the meaningful chunks were emotions.”
In 2017, Heartbeat AI launched its full Enterprise SaaS (software as a service) and API (application programming interface) to help businesses integrate these new applications into their operations. Its first clients came on board after Heartbeat AI’s prototype won the 2016 Insight Innovation Competition for Market Research in Amsterdam.
All companies and organizations have a vested interest in knowing how people feel when they look at or consume a product. For example, an insurance company currently working with Heartbeat AI is using empathetic questions and comments developed by Heartbeat AI to improve its customer service.
As Novikova explains, technology typically has a high IQ and a low EQ (emotional quotient). “We need to put humans at the centre of AI,” she says. Her focus is on a benevolent AI – how applications can use empathy to support users.
When training their algorithm, (Novikova calls it “training the baby”), Heartbeat AI’s method has been full supervision: psychologists and psycholinguists assigning meaning and emotional categories to over 20,000 words and phrases.
“AI absorbs what we put in it,” she says. This technique ensures Heartbeat’s analyses are unbiased.
The future for Heartbeat AI is to test deep learning algorithms that will integrate contextual meanings of words and phrases, and to help the algorithm learn the connections between emotions and needs or motivations.
Novikova, who grew up in the then Soviet country of Kyrgyzstan before immigrating to the United States and then to Canada in 1999, hopes to one day see an eventual breakdown of barriers between people and cultures, and she sees empathy as the key.
“When you see how other people feel, you’re more likely to build compassion rather than to judge.”