As businesses experiment with AI everywhere, one unexpected trend is that they’re turning to AI to help their newfound bots better understand human emotions.
It's a field called “emotional AI,” and the technology is expected to see upward momentum, according to a new Enterprise Saas Emerging Tech Research report from PitchBook.
Here’s why: When companies deploy AI assistants for their executives and employees, and AI chatbots as their front-line salespeople and customer service agents, how can they perform well if the AI can’t understand the difference between an angry “What do you mean?” and a confused “What do you mean?”
Emotion AI claims to be a more sophisticated sibling of sentiment analysis, a pre-AI technique that seeks to extract human emotions from text-based interactions, especially social media. Emotion AI is multimodal, using sensors for visual, auditory, and other inputs to try to detect human emotions during interactions.
Major AI cloud providers offer services that give developers access to emotion AI capabilities, such as Microsoft Azure Cognitive Services’ Emotion API or Amazon Web Services’ Rekognition service (the latter of which has been controversial for years).
According to PitchBook, emotional AI delivered as a cloud service isn't new, but the proliferation of bots in the workplace means it has a bigger future than ever before in the business world.
“With the proliferation of AI assistants and fully automated human-machine interactions, emotion AI is expected to enable more human-like interpretations and responses,” Derek Hernandez, PitchBook’s senior analyst for emerging technologies, wrote in the report.
“Cameras and microphones are essential hardware aspects of emotional AI. They can be located individually on a laptop, on a phone, or in a physical space. Wearable hardware is also likely to provide another avenue for emotional AI beyond these devices,” Hernandez told TechCrunch. (So if that customer service chatbot is asking for camera access, that’s probably why.)
A growing number of startups are launching for that purpose, including Uniphore (which has raised $610 million in total, with $400 million to be raised by NEA in 2022), MorphCast, Voicesense, Superceed, Siena AI, audEERING, and Opsis, each of which has raised modest amounts from various VCs, PitchBook estimates.
Of course, emotional AI is a Silicon Valley approach: using technology to solve problems that humans face.
But even if most AI bots eventually have some form of automated empathy, that doesn’t mean this solution will actually work.
In fact, the last time emotion AI got a lot of attention in Silicon Valley was around 2019, when most of the AI/ML world was focused on computer vision rather than generative languages and art. Researchers have had a hard time with this idea. That year, a team of researchers published a meta-review of the research and concluded that human emotions can’t actually be determined from facial movements. In other words, the idea that AI can be taught to detect human emotions by mimicking what other people do (reading faces, body language, tone of voice) is a somewhat flawed assumption.
AI regulations, such as the European Union’s AI law, which bans computer vision emotion detection systems for certain purposes, such as education, could also nip the idea in the bud. (Some state laws, such as Illinois’ BIPA, also prohibit the collection of biometric readings without permission.)
All of this gives us a broader glimpse into this AI-everywhere future that Silicon Valley is currently building like crazy. These AI bots will either try to understand our emotions to perform customer service, sales, HR, and all the other tasks we want humans to assign, or they might not be very good at the tasks that actually require such abilities. Perhaps what we’re looking at is an office full of Siri-level AI bots circa 2023. Which is worse, you might say, than a bot that needs to be managed to guess everyone’s emotions in real time during a meeting?
