Have You Ever Read Your Clients’ Emotions Through the Monitor?

How Emotions Reading APIs Change the Game

Starting from being a research subject, this tool is now a 20-dollar billion industry, says The Guardian. Emotion recognition can now stand as an individual tool for AI marketing with its own benefits, such as…

• Imagine a world where testing products provides accurate and complete results. Emotions reading APIs can make this possible. Disney tested plenty of films using this option, including ”Star Wars: The Force Awakens” and ”Zootopia”, according to Financial Times. Even Coca-Cola and Intel tested how audiences respond to ads.
• Brands can use them as tools to accomplish their goals. Ford, BMW and Kia Motors declared they want to use it to assess driver alertness, says the same source.
• When you know a customer’s emotions, you get access to unique data that allows you to improve user experience.

Some of the Questions Brands Might Ask

There are six basic emotions and the reader has been trained to recognize each of them - joy, surprise, anger, fear, sadness and disgust. We have reactions for all of them and our faces show each one.

Do people react the same?

ScienceDirect also took into consideration the neutral state of mind, and developed a study that led to surprising results. The accuracy of emotions had a spectacular rate of 96%, while the satisfactory classification accuracy was at 73%. The tool was able to recognize the emotions. Yet, researchers admit that the classification accuracy may be influenced by the fact that people may react to less or greater extent when feeling authentic emotions.

How can the database be accurate?

Let’s take an example. Affectiva, the first business that markets “artificial emotional intelligence” works with 7.5 million faces from 87 countries to support their results.

Can people lie? Since recognition relies on algorithms, people can freely “fake” on purpose what they are feeling. However, when you are watching a two-hour long movie, you might find it difficult to fake your reactions and stay relevant.

Limitations of Facial Emotions Readers

The Microsoft Case

There has been a fuss both when Microsoft released its Perceived Emotion Recognition Using the Face API and when the tech giant announced the tool will no longer be available to the audience. The first limitation comes from the brand itself.

It is important to note, however, that facial expressions alone may not necessarily represent the internal states of people.

In June 2022, Microsoft Corp. announced that it limits access to some facial recognition services and even removes others. It also retired the Face API’s ability to identify “attributes such as gender, age, smile, facial hair, hair and makeup”. 

The company also restricted the usage of software mimicking voices through AI.

The Google Case

Alphabet Inc. blocked AI features that analyze emotions, in an initiative to protect itself from cultural insensitivity. 

Google decided to block 13 planned emotions and review existing ones, like sorrow and joy. It decided instead to focus on frowning and smiling, but without mapping them to emotions.

Individual Studies

Some researchers claim that emotion technology might be biased. Why? Lauren Rhue from University of Maryland says that emotion analysis tech assigns more negative emotion to black men’s faces than to white men.

What did others notice?

• Google API mistakenly labeled black faces as gorillaz.
• Nikon S360 cameras identified Asian faces as blinking.
• Some AI programs have difficulties in identifying gender or darker skin faces.