Chatbots attract the turnover of companies, but fall short when consumers know that it is an AI

From Sephora to Amazon to Dominos Pizza – large companies employ chatbots to communicate with millions of customers around the world.

Advertisements

It has been found that these artificial intelligent systems outsource human employees four times at almost zero marginal costs.

However, new research has shown that when consumers realize that they are talking to a bot, they are less likely to make a purchase because they feel that it is less knowledgeable and less empathetic to their needs.

Scroll down for video

New research has shown that when consumers realize that they are talking to a bot, they are less likely to make a purchase because they feel it is less knowledgeable and less empathetic to their needs

New research has shown that when consumers realize that they are talking to a bot, they are less likely to make a purchase because they feel it is less knowledgeable and less empathetic to their needs

Chatbots are artificial artificial intelligence (AI) that have been developed to simplify human interactions with computers – and are now used in customer service.

Advertisements

Many service-based companies have adopted these systems to help with the workload of communicating with thousands and even millions of customers around the world.

The AIs use voice commands or text chat, have almost zero marginal costs, and appear to outsource most human employees four times.

However, it has created one question: why are these robots not being used more often?

From Sephora (photo) to Amazon to Dominos Pizza - large companies employ chatbots to serve customers

From Sephora (photo) to Amazon to Dominos Pizza - large companies employ chatbots to serve customers

Machine customer conversations, however, reduce purchase rates by more than 79.7 percent when the human consumer discovers they are talking with a chatbot

Machine customer conversations, however, reduce purchase rates by more than 79.7 percent when the human consumer discovers they are talking with a chatbot

From Sephora (left) to Amazon (right) to Dominos Pizza – large companies employ chatbots to serve customers. Machine customer conversations, however, reduce purchase rates by more than 79.7 percent when the human consumer discovers they are talking with a chatbot

A team of researchers went looking for the answer and discovered that a conversation between machine and customer lowers purchase rates by more than 79.7 percent when the human consumer discovers they are talking with a chatbot.

Advertisements

The survey involved more than 6,200 clients of a financial service provider, randomly assigned to people or chatbots.

The revelation of the bots varied from not telling the consumer at all, to telling at the beginning of the conversation or after the conversation, or telling them after they had bought something.

& # 39; Our findings show that when people don't know about the use of artificial intelligence (AI) chatbots, they are four times more effective at selling products than inexperienced employees, but when customers know that the conversation partner is not a human being they are concise and buy less because they think the bone is less knowledgeable and less empathetic, ”says Xueming Luo, a professor and Charles Gilliland Distinguished Chair at Temple University.

& # 39; Chatbots offer improved technological benefits, lower costs for customers and more consumer well-being (offering the product cheaper because bots save money on labor), & # 39; Luo added.

& # 39; With this data, marketers can target specific customer segments to foster customer confidence in chat bots. & # 39;

Advertisements

Although human consumers may be disabled by current chatbots, scientists are working on technology for AI chatbots to help these systems better understand the links between language and how we feel.

And this could lead to the creation of ever smarter home assistants such as Siri and Alexa, and to lay the foundation for emotionally engaging robots.

A team of researchers has developed an emotionally intelligent bone called the Emotional Chat Machine that can respond to messages from human users with an appropriate response based on their mood (process flow chart shown) - to improve human-machine interaction

A team of researchers has developed an emotionally intelligent bone called the Emotional Chat Machine that can respond to messages from human users with an appropriate response based on their mood (process flow chart shown) - to improve human-machine interaction

A team of researchers has developed an emotionally intelligent bone called the Emotional Chat Machine that can respond to messages from human users with an appropriate response based on their mood (process flow chart shown) – to improve human-machine interaction

EMOTIONAL CHATTING MACHINE

The Emotional Chat Machine (ECM) has been developed by a team of Chinese researchers.

Advertisements

Their goal was to create an AI bot that was able to respond to chat messages with an appropriate emotional response.

Users currently have to manually choose the mood in five options.

But in the future, AI software could learn how to select the right emotional response for themselves.

The researchers discovered that more than 61 percent of the people who tested the system preferred the responses to those of a traditional chatbot.

The Emotional Chat Machine (ECM) has been developed by a team of Chinese researchers at Tsinghua University in Beijing and the University of Illinois.

Advertisements

Their goal was to create an AI bot that is able to respond with answers that are not only relevant and grammatically accurate, but also express an appropriate emotional response.

Users currently have to manually select the mood they are in from five options such as happy, sad, disgusting, and angry.

But in the future, AI software could learn how to select the right emotional response for themselves.

And the researchers discovered that 61 percent of the people who tested the system preferred the responses to those of a traditional chatbot.

Speak to the Guardian, Professor Björn Schuller, a computer scientist at Imperial College London, who was not involved with the newspaper, said: & # 39; This will be the next generation of intelligence that will be met earlier in the daily experience, rather than later.

& # 39; It is not a question of whether they are desirable – they are clear – but in which applications they make sense and where not. & # 39;

Chatbots are AI programs that help users communicate with companies and other organizations.

The chatbot studied a data set with millions of real social media interactions. Each was labeled as one of the six emotional categories - anger, disgust, happiness such as, sadness and other (photo)

The chatbot studied a data set with millions of real social media interactions. Each was labeled as one of the six emotional categories - anger, disgust, happiness such as, sadness and other (photo)

The chatbot studied a data set with millions of real social media interactions. Each was labeled as one of the six emotional categories – anger, disgust, happiness such as, sadness and other (photo)

HOW ECM LEARNS

ECM studied a dataset with millions of real social media interactions.

Advertisements

By learning the type of language associated with each of these categories, ECM was then able to assign this emotional content in real-time conversations.

Given the input & # 39; Worst day ever. I arrived late due to traffic & # 39 ;, a traditional chatbot & # 39; You were late & # 39 ;.

But with the help of her emotional intelligence, ECM provided the following answers: & # 39; I am always here to support you & # 39; (such as); & # 39; Keep smiling! It gets better & # 39; (thankfully); & # 39; It's depressing & # 39; (sad); & # 39; Sometimes life just sucks & # 39; (disgusted); & # 39; The traffic is too bad & # 39; (angry)

They can perform automated one-to-one tasks, including customer service and interactive games.

By teaching ECM about the emotional content of messages and their responses, it can properly adjust its own responses.

Advertisements

To learn how to do this, the chatbot studied a data set with millions of real social media interactions.

These received an emotional label based on a sample of 23,000 sentences collected from Weibo, a Chinese social media site.

Each sentence was manually labeled as one of the eight emotional categories: anger, disgust, fear, happiness, such as, sadness, surprise, and others.

Fear and surprise were ignored because they had too few answers, but the remaining labels were applied to the data set.

By learning the type of language associated with each of these categories, ECM was then able to assign this emotional content in real-time conversations.

Advertisements

Given the input & # 39; Worst day ever. I arrived late due to traffic & # 39 ;, a traditional chatbot & # 39; You were late & # 39 ;.

Advertisements

But with the help of her emotional intelligence, ECM provided the following answers: & # 39; I am always here to support you & # 39; (such as); & # 39; Keep smiling! It gets better & # 39; (thankfully); & # 39; It's depressing & # 39; (sad); & # 39; Sometimes life just sucks & # 39; (disgusted); & # 39; The traffic is too bad & # 39; (angry)

TRAINING DATA SET
Emotion Answers
anger 234,635
Disgust 689,295
Luck 306,364
Like it 1,226,954
Sadness 537,028
different 1,365,371

. (TagsToTranslate) Dailymail (t) sciencetech

- Advertisement -