AI and Emotions – Can AI Understand Emotions? Explanation of AI emotion analysis

Some may wonder if AI will have emotions.

In addition, when it comes to understanding other people’s emotions, it is generally a communication ability unique to humans, and many people may think that it is difficult for AI to do so.

In fact, AI has recently become able to analyze human emotions. It’s called “sentiment analysis AI”.

This time, we will introduce the latest trends in AI emotions and sentiment analysis AI.

Do AIs Have Emotions?

In recent years, AI has become a familiar part of our lives. Therefore, some people may be worried that AI will have emotions and threaten human life. AI that can think with this feeling is also called general-purpose AI.

Our emotions are simply responses to electrical signals in our brains. If we can elucidate the mechanism that generates this emotion, it is believed that it is fully possible for AI to have emotions. However, there is currently no general-purpose AI with emotions.

Therefore, it can be said that currently AI does not have emotions. However, as research progresses, it is quite possible that someday AI will have emotions.

Can AI understand human emotions?

In the near future, it is predicted that an era will come when humans and AI will communicate. However, it is important to understand the feelings of the other party in communication. Many people are worried about whether AI can understand human emotions. In the past, it was thought that only humans could grasp and analyze human emotions.

Not all AIs can understand human emotions, but AIs that become human companions or help people in need have the ability to understand emotions.

What is sentiment analysis AI?

Emotion analysis AI refers to AI that can read human emotions, emotional states and changes. The objects of analysis are various, such as sentences, facial expressions, and voices. It is believed that these functions can be used in the service industry, such as customer service, and in marketing.

Ability to respond to other people’s emotions

If AI can analyze emotions, it will be possible to respond according to the emotions of the other party.

Responding according to the other person’s emotions means, for example, not talking to an angry person or cheering up a depressed person. Everyone has experiences like this.

If it is between humans, it is possible to act in consideration of the feelings of the other person as described above, but in the case of AI, there is a possibility that even if the other person is sad, it will do something that makes the other person sad.

Emotion analysis AI enables smooth and natural communication that does not make the other person uncomfortable.

Furthermore, if AI can analyze the emotions of the other person, it will be possible to respond in a way that is close to the other person’s feelings.

The ability of humans to understand the feelings of others and communicate appropriately is called EQ (Emotional Intelligence Quotient), or “Emotional Intelligence Quotient”.

EQ is an ability that is especially required in situations where it is necessary to make the other person feel comfortable, such as dealing with business partners and serving customers at stores.

But EQ varies from person to person. That’s where AI comes into play.

By being able to analyze the emotions of others with high accuracy using AI, even people with low EQ will be able to easily understand the feelings of others.

Using AI to compensate for the lack of EQ may become common in future businesses.

How AI sentiment analysis works

We will introduce the mechanism of AI sentiment analysis by dividing it into the following three parts.

1. Analysis from facial expressions
2. Analysis from tone of voice
3. Analysis from context

I will explain each.

Analysis from facial expressions

One way is to analyze facial expressions.

Reading emotions from facial expressions is something that humans usually do through communication.

Advances in face recognition technology have made it possible to apply AI to emotion analysis. Human facial expressions express emotions. Using AI image recognition technology based on eye and mouth movements, facial muscle movements, etc., it analyzes emotions from the other person’s facial expressions.

Furthermore, it is said that the unconscious emotions and thoughts of people can be inferred from eye movements, line of sight, and pupil size.

In addition, using this technology, face analysis based on facial expressions is not limited to simple emotions. It can also detect subtle differences.

In this way, AI will be able to learn patterns of facial expressions, read emotions, and detect small emotional movements from eye movements and line of sight.

Analysis from tone of voice

The second is the method of analyzing from the voice.

The human voice has volume, intonation, and so on. For example, if you are happy, your voice will be bright and high, and if you are angry, your voice will be low and trembling.

This technology will also make it possible to understand the emotions of call center users and operators. Also, if robots and virtual assistants are equipped with this function, it will be useful for natural communication with humans.

Analysis from context

The third is a method of analyzing from the context and sentences.

Guess the writer’s feelings from the content of the sentences and the tendency of frequently used words.

AI takes in the sentences entered by humans through natural language processing, analyzes them, and judges their emotions. Depending on what words, phrases, and expressions are included in the sentence, it is possible to determine what emotions and feelings the person who entered it has.

For example, if there are words such as “satisfied” and “convenient” in the content of the product review, you can see the positive emotions of the user.

By utilizing this technology, it is expected that chatbots will be able to guess the feelings of the other party and respond politely and pleasantly like a human operator. It can also be used to understand the emotions of customers, etc., from received emails.

Application examples of sentiment analysis AI

Sentiment analysis AI is active in various places. Here are three use cases.

1. call center
2. Customer service, negotiation
3. medical, education industry

I will explain each.

call center

One is a call center.

In a call center that responds by telephone, it is necessary to guess the emotions of the other party from information such as “tone of voice” and “style of speaking.” And some people say, “I’m quick to yell,” while others say, “On the outside, they seem calm, but inside they’re very angry.” Over the phone, it can be said that this decision is very difficult.

AI sentiment analysis can be used to learn from past data, discriminate these, and quantify customer anger levels and satisfaction.

It is also used for the purpose of understanding operator stress. Call center operators had a high turnover rate, which was considered a problem. By using emotion analysis AI, it will be possible to follow up with operators who have accumulated stress, which will lead to the creation of an environment where operators can work comfortably.

Customer service, negotiation

The second is customer service and business negotiations.

Recently, the number of remote customer service and business negotiations has increased due to the corona crisis. However, compared to face-to-face business negotiations, it will be difficult to gauge the emotions of the other party during business negotiations through a camera.

Therefore, using emotion analysis, AI analyzes the “amplitude and interval of voice waveform” and “frequency” uttered by customers during customer service and business negotiations using a unique calculation method and measurement, and visualizes the emotional state of the speaker. To do. The other person’s emotions are not only simple things such as fun and anger, but also important in business negotiations, such as “Do you have a willingness to buy?”

Also, you will be able to respond flexibly according to the feelings of the other party.

medical, education industry

The third is the medical and educational industries.

In recent years, the use of sentiment analysis AI has increased in fields such as digital health, healthcare, and medicine. Among them, the field of mental health is expected. In diagnosing mental health, it is also important to look at the facial expressions, voice, and way of speaking of patients with mental illnesses. Therefore, emotion analysis AI is useful when judging from the patient’s facial expression and speaking style in this way.

In addition, the use of sentiment analysis AI is also being considered in education. Even when you are studying, your expression will change when you are solving a difficult problem or studying your favorite subject. Therefore, if the use of emotion analysis AI becomes more widespread in schools, it may be possible for the AI ​​to understand students’ feelings and propose efficient ways to study, or support them in increasing their motivation. .

Sentiment Analysis Service Summary

There are also services that utilize sentiment analysis AI. Here are five sentiment analysis services.

1. Affdex
2. Empath
3. user local
4. IBM Watson Tone Analyzer
5. Face Targeting AD

I will explain each.

Affdex

This is an emotion analysis service that utilizes image recognition technology.

It is possible to analyze the facial expression of the target person in real time from the existing web camera and convert the emotion into data.

Affectiva, the company that developed it, has about 7 billion emotional data from more than 87 countries around the world, and is characterized by its ability to perform extremely accurate analysis.

Empath

It uses voice recognition technology to analyze emotions from your voice.

Since it analyzes the physical characteristics of voice, it can be used regardless of language.

Also, it can be used not only online but also offline.

user local

Emotions are analyzed from the content of sentences using natural language processing.

It should be easy to understand because you can access the web page and use it directly on the spot.

IBM Watson Tone Analyzer

You can analyze the writer’s emotion and stylistic tone from the text.

Emotions such as anger, anxiety, joy, and sadness can be digitized, and stylistic tones such as confident, analytical, and ambiguous can be digitized.

It can be used in a wide range of ways, such as objectively grasping the customer’s mood from the email they receive.

Face Targeting AD

A service that analyzes the face of a person standing in front of a mirror and distributes advertisements according to gender, age, and emotion. For example, if it can be read that the user is tired from their expression, an advertisement for an energy drink will be displayed.

You can achieve high conversions because ou can hit ads that are close to each person.

summary

Emotion analysis can predict with high accuracy by analyzing human expressions, voices, sentences, etc. with AI.

In business, there are many situations where it is important to understand the feelings of the other party. At such times, sentiment analysis AI may contribute by supporting better decision-making.