The way we deliver customer support is rapidly changing, and one of the technologies at the forefront of that change is Artificial Intelligence (AI).
CEO Jeff Erhardt says “For the purposes of this discussion, we can consider AI, in its essence, the field of computer science that teaches computers to mimic human decision making. This is important, as it underlies our belief that in customer support processes, humans are still the source of intelligence, creativity, and empathy. ”
But how are people using AI in customer support, and what’s in store for the future? Here are a few thoughts from Erhardt about what AI and machine learning can bring to the table and what to expect in the support environment.
Q: What are the layers of AI and how do they affect customer service?
A: Like many things, AI is a popular and often misused term. If I were to distill AI down to its essence, it's broadly the field of computer science that teachers computers to mimic human decision-making. You’ll notice that I didn’t say “think,” or “invent,” or “create.” Those are very different things and certainly researchers aspire to evolve AI into these forms someday. In my view, the field of AI today is nowhere close to that sophisticated. Instead, I think of AI as mimicking human decision-making.
Within that, I might distinguish two different forms of AI which are quite popular. One of them is machine learning, and that’s our field of expertise at Wise. As a company, we focus on something called supervised machine learning, which is basically teaching computers to match results with past behaviors. I’ll contrast that to something that we hear a lot about, which is the world of chatbots, also known as the conversational side of AI.
Many people have heard of the Turing test. The basis of this test is this: can you teach a computer to make a human on the other side of a wall think that it’s actually a person? No computer has passed this test to date, and experts don’t see it happening for quite a while. That said, chatbots are widely considered somewhat risky in the customer support world.
Q: How do you feel about chatbots as your first customer contact?
A: Chatbots are certainly a popular subject of conversation, but the concept of a chatbot is constantly changing. Five years ago, for instance, there was a lot of buzz about big data. Now people ask me what big data is, and I say I have no idea. Big data means several things now and there are different technologies that can be used.
That said, whether we’re talking Big Data or chatbots, I think people must ask themselves, “what am I trying to accomplish?” From a customer service perspective, it’s important to understand what customers fundamentally want.
From my perspective, they want one of two things: either a) an immediate and accurate answer to whatever problem they are having, or b) they want someone to talk to. Those are very different things, and trying to apply any given technology to a given problem without understanding your ultimate goals is a recipe for failure.
Q: Some people consider AI as helpful in collecting information but annoying to customers. How do you know where to draw the line between using AI to resolve an issue versus an handing off the ticket to a human agent?
A: We let our customers draw that line. When it comes to a well-integrated contact center, we often talk about the importance of a customer support technology stack. These approaches shouldn’t be isolated customer support tools, but rather integrated together. Our view is that intelligent systems should seamlessly integrate with the other pieces of the stack, and in particular, integrate with the customer support agents themselves.
Where you choose to draw that line is up to you. Everybody has their own loss function. What does it mean to be wrong relative to how good a prediction is coming out of the system? What do you consider a critical issue? What is a noncritical issue? It’s important to determine where you set that confidence threshold. I don’t think that’s something you should allow someone else to control for you.
Q: Where does AI sit in the technology stack?
A: I would say it can effectively sit anywhere, it’s especially useful in the system view. Think of your business flow. Your customer has a problem and you have an inquiry. You can submit that inquiry through any interface you choose. The question, then, is what do you do with that? How do you take that inquiry and get it to the right account? There’s an AI aspect to that solution. That is, machine learning can help direct that inquiry through multiple channels, be it an OnDemand channel, a human agent, or an automated agent.
Q: What is your evaluation of routing engines within a CRM, and how would they work together to support the routing ecosystem?
A: At Wise, we integrate directly on top of things like Salesforce. In fact, routing is half our product line. We are replacing the need to write and maintain static business rules, and enabling people to answer questions within their existing organizations. At Wise, we like to think of it as inferred knowledge and dynamic rules that are automatically created based on what was done properly in the past. Routing is all about the convergence of technologies to improve the entire support technology stack.
Q: When it comes down to it, can AI be trusted?
A: The short answer is no. Should you blindly trust any technology? The answer is no. The example in the news was Tay from Microsoft, one of the most sophisticated AI companies in the world. They turned their chatbot loose in an unsupervised way. They gave it full trust, and within 24 hours, it turned into a — and I don't remember what words were used exactly — a nazi racist bigot. They took it down, and it was a PR disaster. And so, again, I think the question everyone should answer is: “Would I trust my brand, my company, my reputation to something like that?”
The question is how then should you use these technologies in a way that is beneficial to your business and in a way you can trust and control? That’s something we at Wise spend a lot of time thinking about; how can we put the control back in your hands and not in a black box someplace else?
Still, when it comes to intelligent systems, I encourage you not to be afraid of them. In the world we live in, there’s a lot of talk about machines taking people’s jobs and the social impact of that possibility. It’s something we’ve done a lot of thinking and writing on. Like many other innovations that have come before, people have moved on and figured out how to be creative and innovative. As support leaders, people should think about the same thing for their organizations.
The most fascinating thing for me is taking some pretty sophisticated client organizations to the next level. In some cases, we’ve taken upwards of 50% of support volume off the table. Customer support leaders and analysts might presume that since we’ve saved client organizations time and money, that they would effectively downsize their teams. Instead, our client organizations reassign their customer support staff to higher value tasks. As a result, their customers are happier and their teams are happier because they are able to do something more satisfying, and experience more meaningful work.
Q: What are your overarching goals with Wise?
A: I wanted to change the way these technologies are brought to people. We bring intelligent support solutions to the world by augmenting human experience and human capability. We don’t view it as taking away work, but rather as freeing up human time, to enable people to do what they do well, which is to be creative, to invent, and to be empathetic. The way we do that is to take away the repeatable, the mundane, and the boring work.
To meet with the team at Wise and learn more about how intelligent automation can empower your agents, request a demo of our customer support applications or download our case study .