Five Real Ways Artificial Intelligence Is Upleveling Customer Service
João Graça is the Co-Founder & CTO of Unbabel.
Remember Facebook’s automated personal assistant, M, that was released in a bid to compete with Alexa and Siri? After a series of embarrassing mishaps due to poorly trained algorithms, Facebook abruptly pulled the plug. They weren’t alone; chatbots are infamous for putting their metaphorical feet in their mouths.
While these debacles are tough to watch, the underlying problem is not artificial intelligence (AI) itself. AI succeeds when underpinned with sound strategy and well-trained models. In fact, I’d argue AI is a technology customer service organizations can’t afford to ignore. AI holds massive potential to improve customer service — an area many businesses struggle with.
A 2019 report found that 24% of customer service teams were already using AI and 56% were seeking AI opportunities. And when used wisely, it’s working: Among teams employing AI for customer service, 82% reported increased first contact resolution (FCR) rates, while 79% reported increased CSAT or Net Promoter scores.
As businesses globalize, technology can — nay, must — enable high-quality support that keeps customers happy, no matter their location or language. Here are five ways AI is upleveling customer service:
1. Produce high-quality writing
AI can help customer service agents write more clearly and with better grammar. Improving the quality of writing in customer support channels can increase customers’ perception of a brand. Credibility matters, especially when a customer is already confused or upset.
AI-powered writing assistants enable workers, including customer support agents, to check everything from grammar and spelling to style and tone. This reduces errors while helping agents clearly express meaning to customers.
2. Measure conversations for successful outcomes
Too often, customer support interactions go off the rails because the agent (human or bot) is following a script, rather than focusing on solving the customer’s problem. This may sound simplistic, but let’s take a look at the varied ways customers express intent. A telecom provider might receive one of the following customer messages: “My internet is down” or “I can’t connect to WiFi.”
There are nuances between them, but the goal is likely the same: Get their internet up and running. If an agent is responding to any of these, they must understand what the customer is ultimately after — even if the customer isn’t using plain-spoken language.
AI-powered intent management tools can capture the customer’s goal and measure whether the interaction is moving in the right direction. That’s key to ensuring issues are resolved quickly and to the customer’s satisfaction.
3. Automate FAQs and support ticket triage
Many customer support teams rely on knowledge bases to provide answers to commonly-asked questions. However, keeping FAQs updated and pinpointing the right information at the right time is challenging. Similarly, support ticket triage is key to customer service operations. However, manually triaging tickets to respond efficiently is difficult.
AI can be a huge help on both fronts, helping teams reduce support tickets by continually refining and adding to FAQs. AI-generated templates can simplify knowledge base updates. AI tools can also provide self-service, so customers can get answers to straightforward questions without interacting with an agent.
Furthermore, these tools can automatically triage incoming requests so teams know which tickets to respond to first. Finally, AI can offer suggested or even automated replies, integrate with common workflows and use contextual clues to resolve issues faster.
4. Detect (and reflect) tone and sentiment
Humans naturally excel at detecting tone and sentiment, which isn’t always the case for machines.
Let’s say a customer writes a review that states, “This is the best laptop bag ever. It is so good that within two months of use, it is worthy of being used as a grocery bag.”
Traditionally, an algorithm would scan written language for certain keywords that indicate positive intent or emotion. Here, the words “best,” “good” and “worthy” might convince a machine the writer is satisfied. But any human will pick up on the blatant sarcasm. Often, words that might be coded as “positive” or “negative” mean quite the opposite in context.
An AI-powered sentiment engine can use natural language processing to analyze customer text and understand the sentiment. This empowers customer service representatives to respond with insight into how a customer is feeling. Understanding tone and sentiment can help representatives interact with more empathy, prioritize communications and measure customers’ brand perception over time.
5. Translate any language
Previously, a business could get away with only supporting customers who spoke one of the world’s major languages. Today, smart businesses offer customer support in as many languages as possible.
But hiring customer service agents who speak every language is impossible. Enter machine translation, which uses AI to enable seamless communication across any language. Machine translation can help businesses expand to new markets, manage costs and increase customer satisfaction scores.
However, there are nuances. As mentioned in the previous section, tone and sentiment matter, especially across varied cultural contexts. For example, Japanese and German customers expect formality in business communications, while Portuguese customers are generally comfortable with a more informal tone.
And again, context matters. Basic machine translation algorithms simply translate word-by-word, without looking at conversational context. This can lead to frustration and miscommunication and may turn customers off from a brand. It’s important to achieve high-quality, empathetic interactions when using AI to bridge a language barrier.
One solution is human-in-the-loop machine translation. This means machines handle the rote tasks of translation while building in a quality assurance system with native speakers. Humans check computers’ work for accuracy and empathy and feed this information back into the algorithm to refine it. This strategy overcomes many challenges when it comes to machine translation.
The future of AI in customer service
I believe artificial intelligence will continue to expand its capacity to support customer service. Many teams are working on improving natural language processing, decreasing bias in AI and increasing the speed and quality of customer service interactions. That said, there’s no time like the present to start putting some of the amazing AI tools to work for your customer service program.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?