When the conversation turns to AI, there’s often a Sci-Fi novel’s worth of terminology and jargon that the uninitiated reader has to decode. If you’re looking at using automation for service, then here’s a quick guide to the difference between AI, Machine Learning, and Deep Learning.
Artificial Intelligence as a concept has been around since at least the ancient Greeks, who designed some mechanical devices that could be loosely-termed as intelligent. However the term itself is around 60 years old, and the first applicable AI technologies have only just started coming to market in the last few years.
Machine Learning is a more specific subset of AI. It describes machines’ ability to learn from their mistakes and improve over time. A good example of Machine Learning in practice example is the recent Google AI that beat a world champion at Go. The more the AI plays, the better it becomes at spotting patterns and predicting its opponents’ moves.
Deep Learning is a further iteration of machine learning. It describes machine learning algorithms that run on multiple layers, mirroring how our own neurons function. A now common example of deep learning is the way that smart assistants like Alexa or Siri process speech.
Also important is Natural Language Processing. NLP is the ability for a computer program to understand human speech, regardless of slang or dialect. By being able to make sense of written or spoken language in the messy and error-filled ways humans normally express it, AI capabilities become much more applicable to everyday life.
What does this mean for service? Artificial intelligence and intelligent automation can take over existing tasks and create new efficiencies that your organization couldn’t dream of previously. Machine Learning is just one example. By suggesting responses agents can use to common customer queries, a partially-automated system could learn the most effective replies and language for your customer base. Deep learning capabilities should extend to IVR trees, and put an end to the common “Sorry, I didn’t get” response from many systems that currently rely on processing speech. And NLP is crucial for chatbots, and for analytics that look at all of the conversations your agents have across chat, social, and any other text-driven medium.
It’s important to build a solid understanding of these exciting technologies as they become more prevalent and relevant to the service and customer experience sphere. To learn more, listen to our webinar with Solvvy: The Truth About Bots and Intelligent Automation.
84% of the attendees of our recent webinar, The Truth About Bots and Intelligent Automation, consider Customer Experience Automation a priority for their strategy going forward. What options are available for the automation-minded company, and will a bot deliver amazing service AND make you breakfast? Well, not quite. We got to the bottom of these questions on air, and you can too from the recap below.
Peter Johnson, Kustomer’s VP of Product, and Kaan Ersun, Solvvy’s SVP of Marketing, are both authorities on bots, automation, and using intelligent technologies for better service and support. They discussed the pros and cons of the solutions out there, and made some suggestions for picking and enabling more intelligent service.
Is Automation a Priority?
To kick things off, we started with a poll to take the temperature of the audience. We asked, “Is adopting a customer experience automation solution, such as bots, a priority for you https://s29093.pcdn.co/ your organization?” The results were surprising—the majority of respondents were actively pursuing an automation strategy. Here’s the breakdown:
Yes, this year: 40%
Yes, next year: 32%
Yes, within 2 years: 12%
Not a priority: 16%
Terms You Need to Know
To level-set, PJ and Kaan laid out an overview of the terminology they’ll be using when discussing this complicated technology.
While “intelligent” technologies have existed since Roman times, the term “Artificial Intelligence” came into use in the 1950s—though truly intelligent products just started becoming widely available over the last handful of years. Machine Learning is a more specific application, referring to the ability of machines to advance their program and “learn” from their mistakes without additional programming. A good example is the recent Google AI that beat a world champion at Go. Deep Learning is an even more advanced subset, describing computers that use algorithms that mimic the neural networks of the human brain—meaning they can learn on multiple levels without human supervision.
Bots—Are They All They’re Cracked Up to Be?
But how are these advancements being used on a practical level today? Bots are already taking on a variety of service and service-adjacent tasks within the enterprise, from Digital Marketing and DIY Service, to use cases involving virtual assistants. However, these experiments are still in their early stages. While they may help scale your service, they require a lot of effort to build, and lack customer understanding and the ability to deliver a quality, memorable experience. When you look at the cost and effort to build one versus the level of experience they provide, the math is a bit off.
As PJ put it:
“You’ve probably contacted or been contacted by a support system that tried to act like a human being, but clearly is not. One of our best practices is not to try to seem human, because it can really hurt your brand image and experience.”
On top of that, they aren’t exactly plug-and-play. Service teams have to create replies for every possible input, and they need to be customized for the relevant terminology and details of your business. Actually integrating them with your existing data systems can be a headache, plus they need ongoing maintenance every time you add a new feature or product.
Who’s Using Bots?
Bots may not be the tech overlords they’ve been billed as (yet), but other applications of automation and intelligent systems can supe up your support. And, it’s probably not too late. In our second poll, we learned that most attendees haven’t started using bots yet:
We asked, “What has been your experience with traditional bot technology in your CX operations?” and these were the results:
We use a bot today and love it: 9%
We use a bot today and have encountered some issues: 12%
We use a bot today and have encountered many issues: 12%
I don’t use any bot technology today: 67%
From Bots to Conversational Experience
Before you start experimenting with bots, it’s good to know your options. As Kaan recommended:
“It’s key to have an overarching, holistic automation strategy first—then you can deploy bots as point solutions.”
Bots are a part of this strategy, but not the only focus. Instead, you can also use automation in conjunction with other integrations and platforms to create a stronger experience. Conversational forms look like a chat, but can be used to gather customer info and issues before handing off to a more capable agent to handle the issue. A system that automatically suggests responses to agents works the same as a bot, but uses the added layer of human oversight to learn the right way to respond by tracking your agents’ decisions. And automation is useful for suggesting tags, categorization, macros, helpbase articles, and assisting workflow and reporting—all things that can speed up your experience and make it more efficient, without directly interacting with customers.
As PJ summarized: “Automation is not just about helping the customer, it’s about helping your support organization scale, and identifying areas the product team can improve.”
Kutomer and Solvvy work together to make conversational experience a reality. If you submit a question to Solvvy and can’t find the answer, you can choose to instantly open up a chat in Kustomer and get the answer. Kustomer’s conversational form then collects your personal information, then connects you with an agent who knows your whole customer history.
Where to Begin?
Where do you start the process of using automation or bots strategy if you haven’t already? Kaan had some advice: “Number one, define a strategy, and figure out where the bot can be useful to you, where it won’t work, then pursue new opportunities. Start with the big picture, then move towards implementation.”
Adding to Kaan’s advice, PJ suggested going straight to the data: “First thing: Look at your reporting, and see where you have the highest level of support volume. Look for patterns, see which questions your customers are asking, and what the most repetitive tasks are for your agents?”
If you’re taking a wide-angle approach and carefully planning your strategy, instead of leaping head first into messing around with a bot, your initiatives are much more likely to be a success.
You can always watch the recording HERE, and for more great insights into service, experience, and technology, follow Solvvy and Kustomer.