In 2023, love artificial intelligence—technology that can sense and interact with human emotions—will become one of the mainstream applications of machine learning. For example, Hume AI, founded by Alan Cowen, a former Google researcher, is developing tools to measure emotions from verbal expressions, facial expressions, and tone of voice. Swedish company Smart Eyes recently acquired Affectiva, a subsidiary of the MIT Media Lab that developed the SoundNet neural network, an algorithm that classifies emotions such as anger from audio samples in less than 1 hour. 2 sec. Even video platform Zoom is introducing Zoom IQ, a feature that will soon provide users with real-time emotional analysis and engagement during a virtual meeting.
In 2023, technology companies will launch advanced chatbots that can closely simulate human emotions to create more empathetic connections with users in the banking, education and care sectors. health. Microsoft’s Xiaoice chatbot has been a hit in China, with the average user reported to have chatted with “her” more than 60 times in a month. It also passed the Turing test, with users not recognizing it as a bot for 10 minutes. Analysis from research consulting firm Juniper shows that chatbot interactions in healthcare will grow by nearly 167% from 2018, reaching 2.8 billion interactions annually by 2023. This will free up time. health care workers and has the potential to save approximately $3.7 billion for healthcare systems worldwide.
In 2023, emotional AI will also become commonplace in schools. In Hong Kong, some high schools are already using an artificial intelligence program developed by Find Solutions AI, which can measure microscopic muscle movements in students’ faces and identify a range of negative emotions and feelings. positive. Teachers are using the system to monitor changes in students’ emotions, as well as students’ motivation and focus, so they can intervene early if students lose interest.
The problem is that much of the science-based emotional AI is flawed. Emotional AI algorithms, even when trained on large and diverse data sets, will reduce facial expressions and tone of voice to an emotion without considering the social and cultural context of the person. people and situations. For example, although algorithms can recognize and report that a person is crying, they cannot always accurately deduce the reason and meaning behind those tears. Likewise, a scowling face doesn’t necessarily imply an angry person, but it is the conclusion an algorithm would probably come to. Why? We all regulate our emotional expressions according to social and cultural norms, so our expressions don’t always reflect our inner state. People often do “emotional work” to hide their true feelings, and the way they express them may be a learned response, rather than a spontaneous expression. For example, women are more likely to modify their emotions than men, especially emotions that have negative values ascribed to them, such as anger, because they are expected.
As a result, AI technologies that make assumptions about emotional states are likely to exacerbate gender and racial inequalities in our society. For example, a 2019 UNESCO report pointed to the harmful effects of AI technology’s sexism, with stereotypically designed “feminine” voice assistant systems. and emotionally enslaved.
Facial recognition AI could also perpetuate racial inequality. Analysis from 400 NBA games with two popular emotion-recognition software programs, Face and Face API from Microsoft, shows that on average Black players have more negative emotions, even when they’re smiling. . These results confirm other research showing that Black men must display more positive emotions in the workplace, because they are perceived as aggressive and threatening.
Emotional AI technologies will become more pervasive by 2023, but if left unchallenged and unchecked, they will reinforce systemic racial and gender biases, replicate and reinforce inequalities in the world, while at the same time further discouraging those already marginalized.
