ELI5: emotion AI

Artificial intelligence is a vast area of technology with many different labels, types, subsets and uses. There’s ANI, AGI, machine learning, superintelligence, the list continues. And one of the entries is emotion AI.

As the name suggests, emotion AI combines artificial intelligence and human emotion. But there’s a little more to it than that.

So, what is emotion AI and how does it work? Here’s a simple summary of this subsection of artificial intelligence.

What is emotion AI?

Emotion AI is a subset of artificial intelligence that focuses on detecting and responding to emotion the same way humans can.

It’s also known as ‘artificial emotional intelligence’, ‘emotion recognition’ or ‘emotion detection technology’.

In simple terms, emotion AI is the name for an AI program that can detect and respond to the way you are feeling.

Is it just another name for sentiment analysis?

But wait, isn’t that the same as sentiment analysis? Not quite, though the two are alike. Sentiment analysis analyses text-based input for general sentiment (positive, negative or neutral). Meanwhile, emotion AI goes much more in-depth.

For a start, it’s capable of a wider range of input, alongside text-based communication. (For example, facial expressions and vocal cues in speech.)

It also offers a more specific analysis, telling you the precise emotion behind the communication. So, rather than ‘positive’ or ‘negative’, you get ‘happy’, ‘sad’, ‘angry’, ‘bored’, ‘fearful’, ‘surprised’, and so on.

A good way to think of emotion AI is a bit like a super in-depth sentiment analysis with every bell and whistle possible.

How does emotion AI work?

Emotion AI uses masses of data combined with machine learning to analyse emotional input.

Let’s unpack that a bit. Emotion AI uses a huge catalogue of example data. For instance, photos of faces displaying different emotions, which those emotions pre-labelled. (And the same for vocal recordings, text/keyword recognition and so on.)

When it gets new input for analysis, the AI tool compares it to this catalogue of emotions. The most similar match dictates the emotion it assigns to the new input. This, in turn, informs the AI and adds to the data catalogue. So, it has more examples to use, meaning it ‘learns’ and improves over time.

It then uses the emotions it’s detected to supply an empathetic, emotionally appropriate output.

Where is emotion AI useful?
  • Marketing

Emotion AI could first prove useful in marketing and advertising. Here, it would provide insight into the emotional response that marketing content elicits from consumers. It may then act as a tool to support content targeted at specific emotions.

  • Customer service

Emotion AI will likely shine in customer service, too. Artificial emotional intelligence would enable human-like chatbot support. So, even when agents are unavailable, customers would get an empathetic service. (For instance, during non-operating hours.)

  • Care and mental health

Another commonly cited area in which emotion AI could prove useful is care and mental health treatment. The AI can track emotional responses for signs of anxiety or depression over time. Or, it could provide a human-like companion to help combat loneliness.

The concerns

Most artificial intelligence technology carries with it its share of ethical concerns and questions. Emotion AI is no different.

Artificial emotional intelligence is not as accurate as a human. Emotion and the way that humans convey it is highly complex. For instance, some people shout and swear when they’re angry. Others become very quiet. AI works through generalisations, and so could fall foul of these nuances.

Perhaps more concerning is the suggestion that emotion AI could rob people of genuine human interaction. Emotion AI could become the norm in areas where people would normally get a genuine human connection. (For example, in therapy sessions.) AI might be able to interact emotionally, but it’s still not the same as a real human conversation.

Artificial emotional intelligence

In short, emotion AI is a technology that detects and replicates human emotion and empathy. It’s a potential future for sentiment analysis and could prove useful in many industries.

But for the time being, it’s still young, still learning and still dealing with its share of concerns.