Teaching AI to Recognize Emotion
“A well-accepted theory of psychology, marketing, and other fields is that human language reflects personality, thinking style, social connections, and emotional states. ”
— IBM Research Labs
Last week I found out that IBM was teaching Watson, the famed A.I. that won Jeopardy, to recognize human emotion. They have two tools out now in beta that are quite a bit of fun to try:
The researchers also explained the science behind the tools, for instance:
- The model for Big Five personality characteristics was learned from blogs (Yarkoni, 2010).
- The model for Needs was learned from Twitter.
- The model for Values was learned from forum posts.
It's pretty crazy (and slightly worrisome) to think that a machine is trying to learn human needs by reading Twitter.
But of course I was curious to see what Watson thought about me, so I submitted 4,000 words from my personal blog to the Personality Insights tool (which suggests a minimum of 3,600 works but ideally over 6,000), and this was the result:
"You are sentimental and unaggressive.
You are empathetic: you feel what others feel and are compassionate towards them. You are organized: you feel a strong need for structure in your life. And you are uncompromising: you think it is wrong to take advantage of others to get ahead.
Your choices are driven by a desire for well-being.
You are relatively unconcerned with taking pleasure in life: you prefer activities with a purpose greater than just personal enjoyment. You consider helping others to guide a large part of what you do: you think it is important to take care of the people around you."
Hmmm, complementary? Sure. Accurate? That's up for debate.
The Tone Analyzer is especially interesting in regards to our work at Squarespace (wouldn't it be nice to know if your response to a customer might come off in an unintended tone?!) but I have found its results to be lackluster. The science behind it, however, is fascinating.
"The Tone Analyzer service is based on the theory of psycholinguistics. [...] Psycholinguistics researchers have studied whether the words that we use in day-to-day life are reflective of who we are, how we feel, and how we think.
"It is now well accepted in psychology, marketing, and other fields that language reflects more than just what we want to say. The frequency with which we use certain types of words can provide clues to personality, thinking style, social connections, and emotional states."