No Jitter is part of the Informa Tech Division of Informa PLC

This site is operated by a business or businesses owned by Informa PLC and all copyright resides with them. Informa PLC's registered office is 5 Howick Place, London SW1P 1WG. Registered in England and Wales. Number 8860726.

Lessons Learned from Applying AI in the Customer Service Field

In the rapidly evolving field of customer service, Artificial Intelligence (AI) has become a game-changer, particularly in the contact center. As AI is applied to solve business challenges and improve customer service, we have learned many lessons. This article delves into two key lessons gleaned from the application of AI in contact centers.


Lesson 1: Know Your Audience

The first lesson in applying AI was learning how important it is to know who will be interacting with the AI application. We need to understand who the users of the application are, and what they are looking to accomplish by interacting with the application. A misalignment in expectations here can lead to a world of disappointment and frustration.

The challenge can often be in learning about the audience without being able to interact with them directly. Internal teams and leadership can be great place to start questioning and learning about the audience. However, with teams that have filled their role for a long time, they can often overlook or take for granted some of the audience’s needs.

An example of this happened when we were deploying a voice-enabled AI application to help healthcare users reorder supplies for their equipment. As we deployed the AI application, we discovered that our voice training did not fit part of the audience population – the natural language understanding (NLU) engine was struggling to identify the the user's speech.

After further investigation we learned that this population of users were all wearing mandatory oxygen masks and the speech fed to the NLU was muffled and voices were weak. With some careful work we were able to fine tune the NLU engine to properly work with this new subset of audience.

This was a great teaching moment for us. We now work with new customers to get call recordings and better demographics for the users we are interacting with.


Lesson 2: Knowing How To Listen

Working with teams in the customer service world has been a privilege over the past 20 years and has taught me a lot. As I was recently reading a VuxWorld article, one quote stood out to me.

I think one of the key skills that you get in the contact centres is actually just being able to listen to people. You know, being able to show that empathy.

What a great reminder! One of the greatest forms of empathy we can give another is by listening to them. This second lesson comes in this arena… how do you teach AI how to listen?

As a father of teenagers, I often work with my kids to teach them how to truly listen to someone. This is a task that takes a lot of focus and intention to train them in this art.

How do we take this same skill and apply this deploying AI technology?

Here are a few things to keep in mind.

Who are you speaking with?

  • Do you know the demographic and preferences of the user?
  • Do you know what products/services they use?
  • When was the last time they dealt with your brand?

As we begin to build a profile of who the client is, we begin to change how we listen to them. We can begin to take a few steps in their shoes and be more empathetic towards their needs and how we can best serve them.

In a recent voice-enabled AI application that we built, we had an interesting use case that taught us a lot. For a small subset of the users, responses were often off topic, and the AI engine struggled to keep the conversation on topic.

For example, when we were trying to validate the users’ identity the AI application kept getting responses that were not valid for the questions being asked. Upon further inspection we noted there was something else going on.

The bot would ask the user for their date of birth:

AI Bot: Can you please speak your birthdate?

User: Hey Mom, what is your birthday again?

This was NOT what we expected! This response made the light go on about why we were struggling to validate these users. Here we were dealing with a healthcare advocate and not the user themselves. With some careful conversation design we were able to work around this use case and provide an improved experience for the user. This was a wonderful learning opportunity for us as and taught us to listen just a little more closely!

These key lessons can be an encouragement to each business and developer deploying AI applications. Take a servant hearted approach to deploying AI technology for your users. And remember the greatest form of empathy towards your users is listening to them personally, as well as with your deployed AI technology.

Tim Zarkovacki is writing on behalf of the SCTC, a premier professional organization for independent consultants. Our consultant members are leaders in the industry, able to provide best of breed professional services in a wide array of technologies. Every consultant member commits annually to a strict Code of Ethics, ensuring they work for the client benefit only and do not receive financial compensation from vendors and service providers.