What are chatbots? And how is AI making them better?

AI is making bots more responsive to people

By | 4 minute read | March 12, 2018

Not all chatbots are created equal. Look to AI-powered chatbots.

Our new Martian friend just landed on Earth and is excited to learn about the latest developments in human technology. In this Q&A series, IBM experts explain complicated topics to a Martian (and you).

On Mars, the closest thing to a chatbot is NASA’s Opportunity Rover. And it’s more interested in testing rocks and taking selfies.

I was curious to find out more about chatbots, so I asked Vickie Dorris, an IBM solutions leader in global digital customer care to tell me about chatbots.

Let’s get right to it. What are chatbots?
It’s software that can have conversations with humans (and Martians!). Chatbots allow people to interact with a device or service in the way that’s most comfortable to them—using natural language—rather than traditional methods of interaction.

Why would people use chatbots instead of people?
I prefer to think that bots augment what people do, not necessarily replace them. Bots have a couple of advantages over people:

  • They’re available 24 hours a day, 7 days a week, and in multiple languages
  • Their capabilities work the same regardless of channel or device
  • Once trained, they never forget
  • They don’t mind repeatedly answering the same questions, over and over again

Like “why did you crash your spaceship into my farm?”
Well, more like answering repetitive customer service questions. Though maybe you should fly more carefully…?

Light speed just feels too slow. Would chatbots serve on the front line of customer support?
Yes. In the utility industry, for example, 80% of calls can be attributed to a few key areas, like billing and payments, account management, outages, and service requests. Chatbots would answer the routine customer questions and interactions and then seamlessly deflect more complex questions and issues to staff when needed—freeing people to concentrate on higher value interactions. This lowers the cost of service and improves call center efficiencies and satisfaction.

So chatbots can actually make customers happier?
Absolutely. Customers today are highly demanding. They expect to get answers quickly, at any time, on any channel they choose, so they can move on with their lives.

In a 2018 State of Chatbots report by Drift, the top three frustrations with traditional online experiences were 1) sites are hard to navigate, 2) can’t get answers to simple questions, and 3) basic details about a business are hard to find.

Chatbots are ideally placed to solve these frustrations—providing personalized service anytime, anywhere.

What types of companies could use chatbots?
Many, many companies and governments. Including energy providers worldwide, which are being challenged to revitalize and personalize their customer interactions while controlling costs. And chatbots are certainly not limited to customer service. They’re being implemented internally for front-line HR support, the IT help desk, and for procurement questions.

Field service crews can use chatbots to get answers quickly on site to complete a job as efficiently and safely as possible. To perform an inspection, a field engineer might have to climb a tall mast, which can be dangerous. They could stay hands-free by asking the chatbot about the order of tasks, when the equipment was last inspected, or the status of the previous inspection report.

So where does AI come in?
Not all bots are created equal. Simple bots are basically rules based. If a customer asks, “What’s my account balance and when is my payment due?” a simple chatbot will provide the amount and due date.

AI can also help make chatbot conversations more human-like. That comes in with more sophisticated bots that use artificial intelligence to understand the complexities of language and learn how to engage in more natural conversations that identify and resolve problems.

They can understand a customer’s tone when they are having difficulty making a full payment. The chatbot can empathize, check payment history, and offer a payment extension or plan.

They don’t need to be trained on every permutation of every potential question. They can contextualize human responses to define the underlying meaning of words, especially with non-standard conversational language, such as “My bill needs fixing” or “Gotta bounce” at the end of the chat.

Imagine a million customers asking the same question about bill payments in a million different ways—a chatbot will understand them all.

What role does data play in all these interactions?
All bots need to be trained on data sets specific to what they do. Truly functional bots will be integrated with other systems to improve their capabilities. For energy service providers, these could be the customer, marketing, and outage systems where the chatbot could personalize interactions by accessing a customer’s billing data and completing their payment.

So what’s next for chatbots? Where will they have the most impact?
The early adopter phase is now over. The technology has been proven, especially with customer service. Many organizations are seeing excellent results, so widespread adoption will be next.

Organizations will need to invest more in teams specifically formed to onboard, train, support, and develop bots – such as conversational dialogue designers, UX designers, and business analysts. This technology is moving beyond traditional web and mobile channels into devices like cars, consumer electronics, and conference rooms. This portable bot-based solution will allow consumers to engage with your organization as they transition from their laptop, to phone, to car, and home or office.

Fascinating. The future looks bright for Earthlings! I’ll have to update the Opportunity Rover.
I agree.