How AI is transforming Customer Experience

On May 25, 2018, Toronto Tech Summit held an afternoon summit on innovation focused on artificial intelligence (AI) and customer experience (CX). The summit featured several panel discussions and presentations and talks from booming enterprises and startups.

The anecdote that got everyone nodding their head is the experience of waiting in the phone call for sales or technical customer services and being transferred for a few times before the issue got resolved (or unluckily unresolved sometimes). Like calling to change a phone plan with Rogers or Bell. The pain point from such customer services can be mitigated by the appropriate deployment of AI. This is due to a few factors:

• Most people get used to texting/messaging. How often do you really use your smartphone for calling?
• AI for audio speech recognition and text-based natural language processing has seen a leap in advancement recently.

Some types of customer services could be easily undertaken by AI, such as changing a phone plan to the one that is currently offered by the service provider. For more complicated issue, a human representative might still be needed, but the AI frontline could help the human representative prepare a unique solution to the customer, rather than having the customer re-explain the issue over and over again.

Another interesting idea presented in the summit is distributed contact. If the issue is not time-sensitive and you already have other matters to attend to (such as a crying baby or a unusually work load), who would really like to spend an lump amount of 30 minutes or more on a phone to handle that issue? What if the issue could be handled via your favorite messaging channel (such as WhatsApp or Telegram) over several chunks of time you are available (such as walking to grab coffee or sprinting on a gym bike)?

Emotion is the currency of experience.

The bottom line of customer experience is that emotion is the currency of experience. But current AI technology is still very bad at emotion, compassion, and empathy, at least for the time being. It leaves human with the power to add a touch of humanity in the AI technology that would be deployed. Essentially like garbage-in-garbage-out, an AI model that feeds on racially biased data would yield a racially biased prediction for customer services. It leaves the AI scientists/engineers to undo the bias in the data.

Bias data + Negligent engineering = Bias AI

Speaking from being a customer, I really look forward to a day when I could solve common issues via texting or interacting with the future version of Google Duplex. The end time of waiting in line for a phone call is coming very soon.