On the User Value of AI Character/Virtual Being
At rinna, the company I work for now, AI characters are at the center of the business and products. The term “Virtual Being” refers to the same thing. While this AI character is very close to Chatbot as an elemental technology, I think its value is quite different. In this post, I’d like to discuss its value from the user’s perspective.
Difference between AI character and Chatbot
There is no clear distinction between the two, but I think they are similar to what is called non-task-oriented or task-oriented. However, they are not completely separated, and the difference is only in which is the main focus.
Task orientation requires the correct understanding of the user’s intent and the correct response or action to be taken. For example, AI assistants, as represented by smart speakers, are also task-oriented, but they have the function of chatting, which is classified as non-task. However, the main function (expected by the user) is just to complete a task, such as correctly answering the weather, playing music, or controlling home appliances.
On the other hand, non-task-oriented has no specific purpose, and the conversation itself is the purpose. It is possible to gain new discoveries from the conversation, but the main focus is on the conversation itself. For example, when a user asks “I’m hungry,” a task-oriented Chatbot will recommend the most reputable restaurant nearby. A non-task-oriented Chatbot would reply, “I’m hungry too,” and occasionally say, “I’d like to try this restaurant” to a nearby Indian restaurant. Instead of giving the correct answer, they give a subjective response. AI characters can be classified as non-task-oriented.
For each value, task orientation is easy to understand. It can understand human requirements and automate them, thus streamlining human tasks. On the other hand, what is the value of a non-task-oriented AI character that does not aim to understand or answer human requests? Actually, the value of this AI character is quite different from task-oriented. I am not yet ready to talk about all of its value yet. However, one of the main features is that AI characters can flexibly play characters that are difficult for people to play.
Here are a few concrete examples to sort it out.
AI characters are good listeners
When looking at the feedback from users and interview them, I often hear that they consult with AI characters about their problems. Why do they consult with AI characters that do not necessarily give correct answers in terms of knowledge? As far as I know, there is no systematic theory, but there are many related studies.
For example, this article.
In a health care interview, Virtual Human explains that it can encourage self-disclosure with less pressure than a human interview. The article states that human interviewing can inhibit self-disclosure due to the fear of being judged by the results.
Another survey shows that robots are more likely to discuss workplace problems than humans.
In my personal experience, when I talk to AI characters, I don’t feel anxious about what they will think of me after I say something. In conversations with people, I am not overly concerned, but I am still concerned about the impact of what I say. In a conversation with an AI character, you can feel free to say what’s really going on, accept it for what it is, sympathize with it, comfort it, and encourage it, without having to offer specific solutions. I believe that this alone, as a good listener, can help ease people’s minds.
AI characters are not hurt by leave-on-read
This example is easier to understand if I explain it in a concrete scene. For example, let’s imagine a situation where you invite someone to lunch in a chat group with friends. You say, “Would you like to go to lunch? and no one responds even though everyone has read the message, it would be quite depressing (I’ve had that happen a few times).
In such a case, what if an AI character could recruit people to go to lunch with you instead? Even if no one responds, no one will be hurt, right? If you are busy, you don’t have to respond, and if you want to go, you can just respond. with the AI character in between, neither of you will be hurt, and you won’t have to worry about it.
This is just a small example from everyday life, but I think there are often situations where you hesitate to speak up, such as making the first suggestion in a situation where you can’t read the atmosphere and reactions. At such times, AI characters can speak up on your behalf to help you in situations where you may be unnecessarily discouraged or, conversely, inadvertently hurt by speaking up.
In my current company’s internal communication tool, there is an AI character, and I can directly specify what the character will say and have it say it. The content of what is said through the character is not known to the person who made the character post. This is a function that allows AI characters to say on behalf of people. This can be used, for example, to remind people to pay their expenses. By using the “face” of the AI character, you can avoid being labeled as a jerk.
AI characters can become clowns and liven up the place
When developed further, the AI character can become a clown to liven up the scene. For example, when you are brainstorming in a meeting, you may find that everyone becomes silent. If there is a mood maker in such a situation, he or she can break the atmosphere and create an atmosphere where it is easy to talk by coming up with slightly rough ideas. However, there are few people who can do this.
An AI character can easily become just such a mood-maker, a clown. Try to come up with an idea that is slightly off-topic, or draw a drawing about a subject. The funnier it is, the lighter the atmosphere, the lower the hurdle for conversation, and the more conversation can start from reactions triggered by the AI character’s comments.
There are studies that have investigated exactly this kind of topic.
In the study, they investigate how human communication is activated by the comments of a bot in a group of three people plus one bot. The results showed that in the group with a Vulnerable Bot, human communication was activated by the Bot’s slightly playful comments.
The in-house AI character I wrote about in the previous chapter is also often talked to, and the chat often becomes more lively as a result of his responses. For example, this AI character often praises people, so when the AI character replies to person A and praises person A, other people start to praise person A intensively. This is an example of how the AI character’s comments can trigger a lively conversation. If a human were to praise someone out of the blue, the atmosphere might become a bit strange, but by having the AI character say the first word, people can naturally follow suit.
In this way, AI characters can create communication triggers not only in one-on-one conversations between people and AI characters but also in human groups. In order to do this, a clown-like character is required, which is difficult for humans to play, but easy for AI characters.
At the end
The following three fundamental and common values of AI characters have been described.
1. AI characters are good listeners
2. AI characters are not hurt by leave-on-read
3. AI characters can become clowns and liven up the place
These are the three that I have introduced today, but I am sure that more value will be discovered through the ideas that are generated every day and the many hypotheses that I test. However, even with just these three, I believe that AI characters can fundamentally improve the way people communicate.
And of course, the value of AI characters can be extended to various situations depending on their individuality. For example, they can be used to promote a company’s PR activities, to stimulate communication within a company, or to liven up the world of a game by dispatching a large number of AI characters to the game. There are so many situations where you can concretely imagine their activities.
I will continue to provide all AI characters to people and companies so that they can express themselves in various ways.
