Predictions vary, but expect chatbots in customer service to be mainstream in a few years Gartner Research en Finance Digest. OpenAI has certainly increased the chance of that with ChatGPT. The challenge will soon be to make chatbots appear sufficiently human, without ending up in the uncanny valley. That is the stage where consumers turn away from technology because it becomes too human and creates a feeling of discomfort.

 

Researchers from University of South Florida conducted experiments to determine whether positive emotional behaviors in chatbots improved customer service. The researchers presented their findings in the paper “Bots With Feelings: Should AI Agents Express Positive Emotion in Customer Service?”, in Information Systems Research from December 2022.

Deliver with the customer

Showing empathy, empathizing with the customer, showing understanding: it partly takes the sting out of the conversation with an angry customer. The researchers conducted three experiments to understand the role of AI and emotions in customer service transactions. In addition, the emotional, empathic variant of the AI ​​chatbot was set to use more positive adjectives, such as 'excited', 'happy' or 'happy', including (multiple) exclamation marks.

The first experiment examined whether consumers responded more favorably to positive emotions if they knew that the interlocutor was a bot or a person. Participants were asked to pretend they were seeking help for a missing item in an order. The 155 participants were then randomly assigned to four different scenarios: human agents with neutral emotion, human agents with positive emotion, bots with neutral emotion, and bots with positive emotion. They then asked the participants about the quality of service and overall satisfaction. The results indicated that positive emotion was judged more favorably when the participants were dealing with human agents; if bots showed these positive emotions, it had no effect.

Customer expectations matter

The second experiment investigated to what extent consumers' personal expectations determined the response to the bot. In this scenario, the 88 participants imagined returning a textbook and were randomly assigned to an emotion-positive or emotion-neutral bot. After chatting with the bot, participants were asked to indicate on a scale whether they themselves were community-oriented (social) or exchange-oriented (transaction). If the participant was community-oriented, they were more likely to appreciate the positive-emotional bot, but if they expected the exchange to be purely transactional, the emotional-positive bot worsened their experience.

The final experiment explored why a bot's positive emotion influences customers' emotions. In this round, 177 students were randomly assigned to emotional or non-emotional bots. These results explained why positive bots have less effect than expected: consumers don't expect machines to have emotions, and therefore they can react negatively to emotion in a bot. A 'happy chatbot' can lead to an unhappy or irritated customer.

Conclusions

The results of all experiments show that the use of positive emotion in chatbots is a challenge, especially if companies have no insight into their customer's expectations at the start of the interaction. More generally, the researchers state that the results indicate that the positive effect of expressing positive emotion on a service contact may not be forthcoming if the source of the emotion is not human. In short, think carefully whether you want to make your chatbot (texts) too cheerful and exuberant. (zip tone)

Customer Experience, Featured, Knowledge base