There has long been a chasm between what we perceive artificial intelligence to be and what it can actually do. Our films, literature, and video game representations of “intelligent machines,” depict AI as detached but highly intuitive interfaces. We will find communication re-imagined with emotion AI.

In the midst of a burgeoning AI Renaissance, we’re starting to see higher emotional intelligence from artificial intelligence.

As these artificial systems are being integrated into our commerce, entertainment, and logistics networks, we are witnessing emotional intelligence. These smarter systems have a better understanding of how humans feel and why they feel that way.

The result is a “re-imagining” of how people and businesses can communicate and operate. These smart systems are drastically improving the voice user interface of voice-activated systems in our homes. AI is improving not only facial recognition but changing what is done with that data.

Better Insights into Human Expression

Humans use thousands of subverbal cues when they communicate. The tone of their voice, the speed at which someone speaks– these are all hugely important parts of a conversation but aren’t part of the “raw data” of that conversation.

New systems designed to measure these verbal interactions are now able to look at emotions like anger, fear, sadness, happiness, or surprise based on dozens of metrics related to specific cues and expressions. Algorithms are being trained to evaluate the minutia of speech in relation to one another, building a map of how we read each other in social situations.

Systems are increasingly able to analyze the subtext of language based on the tone, volume, speed, or clarity of what is being said. Not only does this help these systems to identify the gender and age of the speaker better, but they are growing increasingly sophisticated in recognizing when someone is excited, worried, sad, angry, or tired. While real-time integration of these systems is still in development, voice analysis algorithms are better able to identify critical concerns and emotions as they get smarter.

Improving Accuracy in Emotional Artificial Intelligence

Machine learning is the cornerstone of successful artificial intelligence – even more so in the development of emotional AI. These systems need a vast repository of human facial expressions, voices, and interactions to learn how to establish a baseline and then identify shifts from that baseline. More importantly, humans are not static. We don’t all react the same when angry or sad. Colloquialisms don’t just affect the content of language, but its structure and delivery.

For these algorithms to be accurate, they must collect a representative sample from across the globe and from different regions within specific countries. The gathering of a diverse sampling of people presents an extra challenge for developers. It’s your IT developer who is responsible for teaching a machine to think more like a person. At the same time, your developer must account for just how different people are, and how inaccurate people can be in reading each other.

The result of this is a striking uptick in the ability of artificial intelligence to replicate a fundamental human behavior. We have Alexa developers actively working to teach the voice assistant to hold conversations that recognize emotional distress, the US Government using tone detection technology to detect the symptoms and signs of PTSD in active duty soldiers and veterans and increasingly advanced research into the impact of specific physical ailments like Parkinson’s on someone’s voice.

While done at a small scale, it shows that the data behind someone’s outward expression of emotion can be cataloged and used to evaluate their current mood.

artificial intelligence is becoming emotionally intelligent
Communication Re-Imagined with Emotion AI

The Next Step for Businesses and People

What does this mean for business and the people who use these technologies?

Emotional AI systems are being used in a range of different applications, including:

  • Feedback Surveys
  • Coaching
  • Customer Support
  • Sales Enablement

These systems can analyze conversations and provide key insights into the nature and intent of someone’s inquiry based on how they speak and their facial and voice cues during a conversation. Support teams are better able to pinpoint angry customers and take action. Sales teams can analyze transcripts from calls to see where they might have lost a prospect. Human resources can implement smarter, more personalized training and coaching programs to develop their leadership bench.

At the same time, these technologies represent a substantial potential for a leap forward in consumer applications. Voice user interfaces will be able to recognize when someone is sick, sad, angry, or happy and respond accordingly. Kiosks in banks, retailers, and restaurants will be able to interact with customers based not just on the buttons they tap, but the words they speak and the way in which they speak them.

While some of these applications are viable sooner than others, the evolution of artificial intelligence to better understand human emotions through facial and voice cues represents a vast new opportunity in both B2B and consumer-oriented applications.

Rana Gujral

Rana Gujral is an entrepreneur, speaker, investor and the CEO of Behavioral Signals, an enterprise software company that delivers a robust and fast evolving emotion AI engine that introduces emotional intelligence into speech recognition technology. Rana has been awarded the ‘Entrepreneur of the Month’ by CIO Magazine and the ‘US China Pioneer’ Award by IEIE, he has been listed among Top 10 Entrepreneurs to follow in 2017 by Huffington Post.