Are you irked, gratified, satisfied, offended, cheerful? Have you ever thought about how many emotions you experience throughout your day? What if all your devices including your watch, phone, TV, tablet, computer, and digital personal assistant could understand how you feel? Translating and having devices understand sentiment may be the future to personal and professional well-being. Imagine this: “Hey Siri, how am I doing?” Siri: “Frankly, you’ve been sounding stressed lately. Shall we think about a vacation? Southwest has fares for as low as $130 one way to Belize. Shall I look?”
Moodies is an app that analyzes your voice and responds with a characterization of your sentiment. What if we could combine voice and sentiment with fluid interfaces that adapt to your mood? There are many researchers who believe breakthroughs in these areas will be the next progression in personalization.
We are on the verge of Artificial Emotional Intelligence
MIT researchers remind us that AI will only be capable of true commonsense reasoning once it understands emotions. They are discovering that deep learning models can learn subtle representations of language to determine sadness, seriousness, happiness, sarcasm or irony. They are parsing 1.2 billion tweets to train the AI model so it understands emotion. The model tries to predict which emoji would be included with any given sentence. Fun, right? Now all those happy, sad, inquisitive, kissy and frowny faces you use can be matched with words that will eventually evolve into developing AI for emotions. And, it all happens on its own. “This is why machine learning provides a promising approach: instead of explicitly telling the machine how to recognize emotions, we ask the machine to learn from many examples of actual text.”1
MIT’s DeepMoji project will make it possible to develop applications that feel what we say and write, then respond accordingly, making everything more personal and relevant. Even though we are years away from commercial development, we can imagine a world where you will tell your phone: “Find me a flight home,” and your phone app will know where you are, where you’re going and whether you are distressed or relaxed. So will understanding human emotion be the thread to connected dialogue? I ask myself this question every day. I even go as far as exploring the thought of “what is consciousness?” but I slap myself when I get the frowny face telling me don’t go there. Instead, let’s talk about voice.
Enough with the monotone
Do robots including virtual phone assistants have to talk without emotion? I think not, and so does Lyrebird, a digital voice service that replicates your voice using four important elements: resonance, relaxation, rhythm and pacing. Although voice replication is still in its infancy, Lyrebird does an impressive job at creating digital voices from human recordings. They rely on taking voice data sets and feeding them into an AI system that learns in an unsupervised way without any human guidance. Imagine combining feelings with realistic sounding responses. Doesn’t your mind go wild thinking about how we might blend these technologies to design great travel experiences? “Hey Siri, how am I doing?” “I think you’re on to something, Alex.”
1 Iyad Rahwan, Associate Professor. MIT