A Walk Through the Uncanny Valley: How AI is Learning to Recognize Emotions & What This Means for Us
Let’s get one thing clear: Artificial Intelligence (AI) doesn’t feel emotions the way people do. Yet it behaves as though it could. It’s up to you to decide what that means.
AI’s emotional repertoire is expanding. The phrase, “it was written all over her face!” has never been more true. Using only visual analysis, AI systems can predict personal attributes such as political leaning and sexual orientation with brow-raising accuracy. When’s the last time someone looked at you and yelled “LIBERAL!”?
AI systems are also revolutionizing the way we diagnose mental health. We’ve long needed a more precise way to do this — unlike other diseases, depression can’t be diagnosed by physical examination. AI provides new tools for diagnostians: these systems detect indicators of depression and suicide better than their human counterparts by detecting subtleties that typically go unnoticed. AI systems have analyzed 100 million texts from the Crisis Text Line and found that those who use the crying face emoji are more likely to commit suicide than those who use the word suicide. AI is also able to pick up on other indicators of mental health such as vocal patterns and body language in Instagram posts.
AI systems are used in other industries to evaluate emotions. In retail, they’re used to analyze shoppers’ emotions as they stroll through the aisles. Smart shelves, in particular, take note of how long customers stop to look at a product, whether they look happy doing so, and if they return to the shelf. These systems are faster and more accurate than shopper self-reports.
It’s clear how AI helps companies get data from their customers, but what do customers want from AI? Capgemini conducted a study to find out, and the results are telling. 64% of consumers want AI to be more human-like, though they want to know if they’re interacting with a human or a bot. Ultimately, a mix of AI and human is preferable. Capgemini also reports that soon AI systems will be able to make small talk with customers about the weather or upcoming holidays. There’s a strong incentive for companies to make their AI assistants more human-like: customers are likely to spend more, and it boosts company credibility and trust.
The flow and tone of human to human interaction is dictated by the emotions betrayed by the participants. Think of all the ways you can say hello: Hello! Hello? Hello…. Our social standing depends on our ability to read others’ emotions and respond accordingly, allowing us to act more empathetically. Does this ability need to extend to robots? Do we want it to?
Whether we want them to or not, it’s possible that machines’ abilities to assess emotion will soon surpass ours, at least in speed and scale. According to MIT, “Machines can analyze images and pick up subtleties in micro-expressions on humans’ faces that might happen even too fast for a person to recognize.” Machines have speed on their side. Plus, they don’t get tired.
Many consider empathy to be the most essential human quality and perhaps the final frontier for Emotion AI. Emotion conscious AI assistants make sense in some situations more so than others, such as in healthcare or emergency assistance but not when buying shoes. Machines will certainly behave as though they empathize, but it’s up to the user to decide whether that counts as empathy.
Empathy may be our last frontier too. To create machines that empathize, we need to decide what this means to us. Machines will only be as empathetic as their creators.
By Lizzie Ottenstein, New York, NY
Articles from the piece:
How Artificial Intelligence Can Save Your Life
Face-reading AI will be able to detect your politics and IQ, professor says
New AI can guess whether you’re gay or straight from a photograph
Are AI Learning Scenarios Unpredictable Enough?
Keep reading:
How AI Will Make Us Better Risk Takers
