IBM introduced a new feature in its Tone Analyzer service Thursday designed to help chatbots better understand consumers' tones and respond appropriately.
The service detects communication tones in conversations to indicate frustration, satisfaction, excitement, politeness, impoliteness, sadness and sympathy, according to IBM.
By automating certain functions that normally require human labor, chatbots can streamline processes and save businesses time and money. But to make them work well in the call center, the chatbot has to have a basic understanding of sentiment, or risk frustrating customers further. Until now, chatbots have been poorly equipped to recognize sentiment.
But it’s a problem worth addressing. A report released by Juniper Research in May found chatbots have the potential to save businesses in the healthcare and banking sectors more than $8 billion per year by 2022, up from $20 million in 2017. Chatbots are particularly effective in the call center, where Juniper found they can potentially save businesses an average of more than 4 minutes per call when compared with traditional call centers.
New chatbot offerings for the enterprise have been on the rise this year. In March, Amazon Web Services announced Amazon Connect, a self-service, cloud-based contact center service for the enterprise. The service uses Amazon's virtual assistant Alexa to respond to questions over the phone or via text, its Lex chatbot building service and its text-to-speech program Polly. It's not clear if AWS has mastered the "tone and sentiment" issue yet.
Some companies, including Intel, Twitter and IBM, are also experimenting with sentiment-analysis software to determine employee concerns and, in some cases, develop programs to help improve the likelihood employees will stay on the job in the ultra-competitive tech arena.