|You might not always remember what you’ve felt but new AI technologies are learning how to recognize human emotions, wrote Meredith Somers of management school MIT Sloan / Photo by: Rawpixel.com via Shutterstock|
How did you feel the last time you watched a TV commercial? Did it make you laugh or did it make you angry? You might not always remember what you’ve felt but new AI technologies are learning how to recognize human emotions, wrote Meredith Somers of management school MIT Sloan. These technologies will use that knowledge to improve health care, marketing campaigns, and more.
In a post written by Laurence Goasduff in Gartner’s news platform Smarter With Gartner, Annette Zimmermann, research vice president at Gartner said, “By 2022, your personal device will know more about your emotional state than your own family,” as cited by Erin Cunnigham of technology and business magazine BizTech.
What Is Emotion AI?
Also known as artificial emotional intelligence or affective computing, it is a subset of AI that simulates, reacts, measures, and understands human emotions. This field of AI “dates back to at least 1995” when Rosalind Picard, an MIT Media lab professor, published “Affective Computing.”
Research scientist with the Affective Computing Group at the MIT Media Lab Javier Hernandez explained that emotion AI is a tool that facilitates a more natural interaction between humans and machines. He said it is akin to interacting with other people. When we communicate with others, we tend to look at their faces and body and change the way we interact accordingly.
MIT Sloan professor Erik Brynjolfsson noted that machines are good at analyzing massive volumes of data. Machines can listen to voice inflections, recognize them, and correlate those inflections with stress or anger. They can also analyze subtleties in micro-expressions in a person’s face, which might be too fast for a human to recognize.
Brynholfsson added that humans have a lot of neurons in their brains for social interactions. He said, “We’re born with some of those skills, and then we learn more.” For him, it makes sense to take advantage of technology to connect with our social brains. Machines that can speak the language of emotions are going to foster better, more effective interactions with us. Emotion AI was not an option two or three decades ago but now, it’s on the table, Brynholfsson noted.
Industries That Are Using Emotion AI
1. Assistive Services
Some people with autism struggle to communicate emotionally. Hence, AI can be used as “assistive technology,” Hernandez said. For example, wearable monitors can discern subtleties in facial expressions or body language in someone with autism that others might not be able to detect.
2. Call Centers
Cogito, a company co-founded in 2007 by MIT Sloan alumni, has a technology that can help call center agents identify their customers’ moods on the phone, allowing them to adjust how they will converse with their clients in real-time. The company’s voice-analytics software is based on “years of human behavior research to identify voice patterns.”
In 2009, Rana el Kaliouby, Ph.D. and Picard founded Affectiva, a Boston-based emotion AI company. The company focuses on automotive AI and advertising research. With the latter, the client is given access to the company’s technology once they vet and agree to the terms of Affectiva’s use such as promising not to use the technology for lie detection and surveillance. With their consent, the technology uses their phone or laptop camera to capture their reactions while watching an advertisement.
Emotion AI and the Risk of Bias
Emotion AI is prone to bias because emotions are subjective, explained Mark Purdy, John Zealley, and Omaro Maseli of Harvard Business Review, a general management magazine. A study by Lauren Rhue found that emotion AI “assigns more negative emotions” to individuals of certain ethnicities than to others, via journal portal SSRN.
What if emotion AI is used in the workplace? It’s possible that the technology will consistently identify an employee as exhibiting negative emotions, thereby affecting their career progression. The technology is also not sophisticated enough to understand cultural differences in expressing and reading various emotions. This makes it harder for AI to draw accurate conclusions. For example, flashing a smile might mean one thing in Germany while it might mean another in Japan.
Consider a Japanese tourist needing assistance while shopping in a store in Berlin. If the shop uses AI to prioritize which individuals to support, the technology might misinterpret their smile. Since it is a sign of politeness in Japan, the shop assistant might mistake their smile as an indication that they did not need help. Thus, businesses are more likely to make mistakes when they confuse these subtleties. If bias is left unaddressed, it can perpetuate stereotypes and assumptions “at an unprecedented scale.”
Welcoming Emotion AI
Businesses should promote a healthy discussion before applying emotion AI into their operations, Hernandez recommended. Discussions should include the benefits of using emotion AI, as well as the possible uses of this technology and how it will be used in private ways. Brynjolfsson warned that emotion AI should be appropriate for everyone and it should b=not be confined to the subset of the population used for training the algorithm.
Emotion AI is a promising technology but it should be used in a fair manner. Biases should be minimized as much as possible so as not to perpetuate stereotypes. For emotion AI to improve, programmers and developers should also study how people in other countries express their emotions.
|Emotion AI is a promising technology but it should be used in a fair manner. Biases should be minimized as much as possible so as not to perpetuate stereotypes / Photo by: Jo Theera via Shutterstock|