In April, Hume AI, a startup, unveiled a demo of its latest creation: EVI, an emotionally intelligent artificial intelligence voice referred to as the "Empathic Voice Interface. " Engaging with EVI resembles interacting with newer generations of voice assistants powered by AI, such as Siri or Alexa, but with the added capability of ChatGPT. You can request EVI to recite a poem or provide an explanation of the causes of the French Revolution, and in response, a gentle and composed voice will deliver a haiku or initiate a brief lecture on eighteenth-century France. Like other AI voice bots, EVI does experience occasional bugs and delays. What sets EVI apart is the display of nuanced emotional expressions it claims to identify in your voice after you request a poem. The AI presents a list of up to 48 distinct emotional expressions, ranging from awkwardness and confusion to contempt and surprise.

Hume AI asserts that its ability to collect data on a larger scale and recognize a greater variety of subtleties surpasses traditional studies conducted by emotional scientists. Founder of Hume AI, Alan Cowen, explained, "Those 48 emotions go beyond what emotional scientists have historically studied because we can gather data on a much larger scale to train these models, and we can capture a wider range of nuances. " Hume's empathic AI is trained on podcasts, other media content, and recordings from their own psychological experiments. The company also offers AI that analyzes facial expressions, aiming to help businesses and chatbots better understand their customers. The cost of using EVI is approximately $0. 10 per minute, which may vary depending on the client. "We currently have clients in customer service, such as Lawyer. com, and big tech companies utilizing different technologies we've developed, " Cowen noted. Lawyer. com employs Hume's technology to enhance its 1-800 line, as call centers seamlessly align with the capabilities of recognizing human frustration. However, Cowen aspires to achieve more significant goals, aiming for personal AI assistants capable of genuinely comprehending individual needs and optimizing well-being. "It learns from you over time, tailoring itself more closely to your preferences, " Cowen added. Artificial intelligence trained on an individual's voice and facial expressions may one day alert them, saying, "Hey, have you noticed that you often feel tired and irritable around 3 PM every day?" While this may sound helpful, it becomes slightly ironic when a soothing robot voice kindly reminds you that Starbucks Frappuccinos are available at half price until 4 PM. "We are more susceptible to making unnecessary purchases during specific emotional states and certain times of the day, " warned Ben Bland, one of the Institute of Electrical and Electronic Engineers' contributors in developing ethical standards for empathic AI. Bland also raises concerns about AI's potential impact on our emotions, comparing it to the effects smartphones had on our attention spans. "If you have a computer game that adjusts its settings based on estimations of your excitement and enjoyment, you might develop an addiction to the game. Emotional desensitization is also possible, " Bland cautioned. Naturally, all of Bland's worst-case scenarios assume that a robot can accurately infer emotions in a scientifically reliable manner, which remains a topic of debate. "I don't believe these technologies work as effectively as often claimed by those who market them, " stated Andrew McStay, director of the Emotional AI Lab at Bangor University in the UK. McStay referred to studies that have shown existing emotion recognition technology tends to assign more negative emotions to the faces of Black men compared to white men, perpetuating harmful stereotypes. He further emphasized the broader question concerning the true nature of emotions and the extent of variation in emotional expression among individuals and cultures. "There is a problem when these claims about these technologies suggest the existence of a discernible biological program while neglecting the social and cultural aspects, " McStay concluded.