Artificial Intelligence (AI) is an important and evolving concept that is having significant impact within the Customer Experience industry — and it’s a topic that is being talked about on a seemingly daily basis at this point. But is AI really ready for primetime in customer care?
I spoke with Michael Johnston, Lead Inventive Scientist at Interactions, about frequently asked questions about AI and Machine Learning as they apply to customer care.
How do you define Artificial Intelligence and Machine Learning ?
Artificial Intelligence refers to the capability of a machine to imitate intelligent human behavior. Put another way, AI technologies are algorithms that attempt to mimic things that humans do. Machine Learning, on the other hand, is the science and engineering of giving computers the ability to learn without being explicitly programmed — algorithms that learn from data.
AI and Machine Learning are often discussed in conjunction with one another, but it is important to note that not all AI techniques use Machine Learning, and Machine Learning is also used for other things besides AI, such as decoding genetic sequences.
What are some specific examples of AI technologies?
Some examples of AI technologies that are commonly used today include:
Speech Recognition: Taking audio and working out what the words spoken are.
Natural Language Understanding: Taking sequences of words and determining the intended meaning.
Computer Vision: Recognizing objects and understanding the world — to provide sensory input for control of a driverless car, for example.
Dialog Management/Conversational AI: Ability to conduct a natural conversation with a user. Taking in the meaning conveyed by the user, thinking, and deciding what to SAY and DO next.
What is the potential use case for AI in customer service?
One of the key applications of AI is to combine these technologies — speech recognition, natural language understanding, dialog management and so on — to create Intelligent Assistants. Intelligent Assistants are interactive systems that can communicate naturally with humans and assist them in accessing information and completing tasks.
Most of the Intelligent Assistants that people are familiar with today are consumer facing and somewhat general in purpose. Siri, Cortana, and Google Now, for example, are Intelligent Assistants that make it easier for a consumer using a phone, tablet, or other device to get things done. The majority of what you can do through these assistants was already possible through the graphical interface of the device, but the Intelligent Assistant enables a single point of entry to perform a broad range of different tasks.
Much the same argument applies to enterprise care. Customers need to interact with a business to fulfill numerous different types of requests such as account access, billing, sales, troubleshooting, and so on. An enterprise Intelligent Assistant powered by AI technologies can provide a simple and consistent point of entry to access numerous different services that get lost in a complex graphical interface or the structured voice menus that are typical of interactive voice response systems (IVRs).
An enterprise Intelligent Assistant could be something you deploy in the voice/telephony channel, or through web chat, text, mobile applications, or to support social media customer care. The ideal is to have all of these channels supported by the same underlying Conversational AI in order to offer the customer a seamless omni-channel experience. For example, when a customer reaches out through the mobile channel, the Conversational AI should be aware of any recent customer service calls in the telephony channel: “So sorry you were having trouble with your internet service yesterday, what can I help you with today?”.
Are Intelligent Assistants the only application of AI for customer care, or are there other ways that AI technologies can impact the contact center?
Front-end use of AI technologies to enable Intelligent Assistants for customer care is certainly key, but there are many other applications. One that I think is particularly interesting is the application of AI to directly support — rather than replace — contact center agents. Technologies such as natural language understanding and speech recognition can be used live during a customer service interaction with a human agent to look up relevant information and make suggestions about how to respond. AI technologies also have an important role in analytics. They can be used to provide an overview of activities within a call center, in addition to providing valuable business insights from customer activity.
How is using AI in customer care different from other applications?
For one thing, it is possible to be considerably more aggressive in the application of some newer technologies specifically in the case of agent support, since there is still a human agent who can choose whether or not to use the supporting information. For example, there are new techniques applying deep learning technologies (including sequence-to-sequence learning) that can learn how to respond to a customer question from large sets of example dialogues.
The danger of applying this kind of technology to a fully automated Intelligent Assistant is that risk remains that the automated interface will provide a suboptimal or uninterpretable response that could negatively impact the individual customer experience or even the brand as a whole.
In the case of AI for agent support though, critically, you have a human in the loop alongside the AI solution. That way, the human agent can choose to approve or disapprove automated or partially automated responses. By doing so, human agents also help to continuously train and teach the Conversational AI over time.
What types of customer service interactions represent the best opportunities for automation?
As we currently see with most traditional IVRs, automation is often used to route a customer to
the right agent or queue. Moving beyond that, automation can also be used to handle self-service transactions. Here, the best place to start is with high volume transactions that already have the appropriate back office APIs that can be easily integrated with the AI solution. From there, companies should move into areas such as technical support, interactive troubleshooting, and interactive sales.
What recommendations do you have for contact center leaders looking to implement AI?
One piece of advice that I would give is: even though none of these solutions are yet perfect, don’t worry about the performance of an AI solution impacting your customer experience. Even if an automated solution cannot handle everything your customers throw at it, there are human-assisted AI solutions within the marketplace that can ensure that the quality of experience for the customer is maintained by applying live human assist in a variety of ways. This can take the form of either an analyst who provides support to a AI-based speech and language solution as it learns, or AI that is used to directly support customer service agents. Ultimately, humans can support AI, and AI can support humans.
Do you think we will see a tipping point where call centers won’t be necessary anymore?
In the foreseeable future, no. But, I do expect that what agents are doing will continue to shift. More and more, the simpler tasks will be handled through digital channels with various levels of automation, and more complex issues will still require human intervention or supervision. Increasingly, the workload of human agents will be restricted to complex issues and these agents will rely on AI-enabled agent support systems to optimize their performance.