The terms artificial intelligence (AI) and chatbot are becoming increasingly ubiquitous. This is due, in part, to the growing number online customer self-service experiences powered by one or both of these technologies.
Alas, with ubiquity comes misunderstanding. How often do we hear the terms chatbot and AI misused, or used interchangeably? To help clear the air, here’s a closer look at chatbots and AI and their close ties to customer self-service.
A brief introduction to AI
There’s a reason that the work of human beings cannot be entirely automated. Our brains are uniquely suited to naturally learn and adapt based on thousands upon thousands of daily interactions. The “data” come from many places, much of it rooted in human-to-human interactions.
As such, we can empathize with, and adapt, “humanized” behaviors. In particular, our brains can parse natural human language (colloquialisms, for example, or figures of speech) that machines sometimes struggle with.
Still, mimicking this inherent capability of the human mind is the aim of most AI technologies. As SAS puts it, “AI works by combining large amounts of data with fast, iterative processing and intelligent algorithms, allowing the software to learn automatically from patterns or features in the data.”
What results from this process is the ability to execute certain tasks the same way humans would, all while self-learning and improving. This, of course, includes customer service interactions.
Enter the chatbot.
What is a chatbot?
A chatbot is a software program that interacts with humans by way of conversational (text- or voice-based) interface. Usually, chatbots live inside some kind of existing messaging application, and they do their darndest to respond to customers the way a human being might.
Essentially, the work of of chatbots, as Google sees it, is “using data to answer questions.”
It’s how those chatbots interpret and use the data that differentiates their capabilities, use cases, and efficacy. Some chatbots use natural learning processing, for instance, so they can respond to inputs with certain actions, such as serving up content that answers a particular question. Though these chatbots can be quite complex, and use entire catalogs of different “playbooks,” they are typically rules-based at their core—not powered by AI, that is.
Other, more advanced chatbots can be powered by AI. Whereas a rules-based chatbot will, say, parse customer input to scan the knowledge base for articles related to certain words and phrases, an AI-powered chatbot will actively learn from that interaction—and interactions like it—to improve performance during the next interaction.
AI-powered chatbots scan and pull from an almost endless source of data, respond to that data proactively, and self-learn along the way.
AI, chatbots, and self-service
The Gartner Market Guide for Conversational Platforms predicts that, by 2021, 15% of all customer service interactions will be completely handled by AI. There’s a reason behind this tall prediction from Gartner analysts. Chatbots and AI lend very, very well to automated customer self-service. You’ll find these technologies in e-commerce settings and B2B—in travel, real estate, and medicine.
That’s because they work.
What used to require a large amount of overhead and contact center headcount can now be handled with a chatbot. Even simple, rules-based chatbots can handle a high volume of lower-tier, repeat issues that would otherwise be handled by tier one in the contact center.
And this can lead to impressive efficiency gains, both in terms of time and real-dollar savings.
But it is the human being on the other end, first and foremost, for which we build our AI applications and chatbots to closely mimic the human mind. Customers are happier when a customer service chatbot can anticipate demand, decrease perceived effort, and be there to answer questions any time of day.
Still, as the CCW Digital Special Report on Chatbots concludes, “bots are only as good as the intelligence driving their operation.” This, in turn, elevates the importance of knowledge management and, specifically, the content we create for our human customers within the chatbot experience. The better the content is that we feed out chatbots, the better they’ll be at helping customers out.
Let’s take it one step further. The better those chatbots are at at automating simple customer service requests, the more bandwidth agents will have for higher-touch interactions. And those interactions, too, will require access to easily accessible, highly relevant content.
This is just one of the ways that chatbots are changing the ways we deliver our knowledge content.