If you’ve spent any amount of time with Amazon’s Alexa, then you know that she can be a great DJ. When my family recently rented a vacation house that came equipped with two Alexas, we were eager to test her limits and began peppering her with questions. My five-year-old daughter was particularly thrilled that she could command Alexa to play all her favorite songs from Moana and Sing. My husband and I could, at least, ask Alexa to turn the volume down.

We learned pretty quickly what some of Alexa’s limits are. She wasn’t hooked up to the house’s lighting system or to any other appliances, but she could play music and search the Internet. In addition to being able to find and play pretty much any song we could think of, Alexa is good with facts. We learned who the oldest living person on the planet is, as well as her age (117), but were frustrated when Alexa struggled to help us find local restaurants. We also found that Alexa preferred direct commands. Asking a question too politely (“Could you please…”) confused her, and about other things, whether she preferred this or that, she often replied, “I do not have an opinion.” Of course, we knew she would be neutral, but it was fun to see how she was programmed to reply.

It’s a whole new world

Alexa is an example of artificial general intelligence (AI), meaning that she can do or know many things (much like Siri or Watson or Bixby), but cannot self-learn and does not have emotions or opinions or preferences.

What I noticed about myself, aside from being able to enjoy a song on-demand, without having to walk over to the stereo, is that it felt strange to speak so sternly to a machine that had a name and a female voice. And listening to my daughter telling Alexa what to do, or repeating herself in different ways until Alexa heard her, or until the command registered, gave me a sense of how I probably sound, at times, speaking to my daughter.

What I noticed about myself, aside from being able to enjoy a song on-demand, without having to walk over to the stereo, is that it felt strange to speak so sternly to a machine that had a name and a female voice.

The comparison to a toddler or young child is apt. Alexa, as well as most AI-powered technology currently in use today, is in its infancy. Or—not quite its infancy. According to Moore’s Law, computing power doubles every 18-24 months. So even though artificial intelligence might feel new, it’s not. It’s just finally getting good, and today we’re seeing the market beginning to flood with AI- and robotics-powered proof of concept products and services, affecting nearly every industry, all around the globe.

The future has arrived

Some say we’ve entered the Fourth Industrial Revolution, and even if we welcome this time as a greenfield of opportunity, there’s also reason for concern. A study from Oxford University reported that 47 percent of U.S. jobs are at risk due to automation within the next 20 years. The reality, as AI and robotics iterate on existing products and services, is that the way we interact with objects, with businesses, A lot.

We worry: Which jobs will be lost, and how can we prepare an entire population to pivot? And how soon? And, of course, which products will we adopt (maybe helpful airport robots?) and which will fall to the wayside? (I’d put money on the expensive, hands-free suitcases.) If nearly everything can be made “smart”… what should be?

If nearly everything can be made “smart” … what should be?

Even if Alexa still seems a long way from being truly useful in the home, the reality is that some products have reached a stage of maturity where the AI can begin to take on some of our cognitive work better and more consistently than we humans can. Consider Tesla’s self-driving cars, which, as of last October, can operate “at a safety level substantially greater than that of a human driver.” It seems unlikely I’ll ever be teaching my daughter how to drive.

We’re only human

Depending on how we look at technology advancements and the relational changes they imply, we’re presented with a glass-half-empty or half-full scenario. At the National Retail Federation’s 2017 BIG Show this past January, Kate Ancketill, CEO and Founder of GDR Creative Intelligence, explained there are valid fears around job displacement and shifting social demographics. At the same time, there’s also the possibility that we’re entering an era of what she called “radical abundance.” As intelligent machinery picks up more of our rote, repetitive work, we humans become free to tackle more challenging, meaningful work. We can accomplish more, go further, dream bigger.

That technology can rival human performance is a scary notion, but can also be comforting. A chatbot or robot will never feel the effects of having stayed up too late binge-watching TV, fighting with a girlfriend, or undergoing a juice cleanse—

A chatbot or robot will never feel the effects of having stayed up too late binge-watching TV, fighting with a girlfriend, or undergoing a juice cleanse.

Instead, technology can bring reliability, efficiency, and gains in productivity. As we increase the ways we interact with AI and with technology, there’s also ample opportunities to increase the ways in which we are distinctly human, and to create stronger human interactions and to focus on building community. One of the first ways businesses can do this, Anketill said, was to “use these technologies to deliver what customers want.” She urged companies to “think about whether any aspect of your customer service can be improved with AI.”

Fast service, with a personal touch

Fortunately, what customers want hasn’t changed so dramatically. The best customer service, say customers, is still fast, easy, and personalized. The difference today, according to research from Loudhouse, is that customers are less patient and

One reason for this shift in expectations is that messaging apps are opening up as new, preferred channels for customer service. It’s easy for consumers to send a private message through Facebook Messenger, Twitter, or SMS, and they expect a response from a brand much like they’re used to from a friend through those same channels. In fact, Loudhouse found that consumers expected a response within 10 minutes—which is a great deal faster than the two hours allowed for more public forms of social media outreach like tweeting at a brand.

This shift in channel outreach creates a ripe opportunity to leverage AI and chatbots to help take care of easily automated tasks, such as requesting basic information from a customer or routing a question to the right person. With AI, companies can meet customers where they are, but can also scale upticks in volume. Even better, the more chats a bot handles, the more accurate the bot becomes.

Even so, Loudhouse reported that 86 percent of respondents said that being treated like a person was more important than benefitting from advances in smart customer service technology. It might seem like consumers want it both ways, but it makes sense. The disconnect I felt with Alexa, considering how forcefully I needed to speak to “her” and how it made me feel (confused, empathetic toward something I know doesn’t have feelings), took away from the technology’s cool factor.

Enter the “customer service centaur”

As Ancketill ran through example after example of intelligent technology already in use—including, but not limited to smart ice cubes that tell the bartender when you need another drink, and smart locks that allow grocery deliveries to be brought into your house while you’re not home, to a beer company that adjusts its recipes based on customer feedback submitted via a Facebook Messenger chatbot)—she kept coming back to these two points: We have to embrace technology in the service of meeting customer needs (her words: “You have no choice”), and we also have to continue investing in people.

Ancketill introduced the concept of “service centaurs,” inviting an image of the half-horse, half-human creatures from Greek mythology who tended toward wild, unruly behavior. It’s less a direct metaphor and more the idea that the future of customer service looks increasingly like a hybrid of smart, automated technology, while also becoming more human than it is today.

What does that mean? Customer service personnel will need to be “enhanced employees”—deeply, empathic human creatures capable of nuanced emotional intelligence. “Empathy is the one thing that I hope and believe will remain a human ability,” Ancketill said.

Customer service personnel will need to be “enhanced employees”—deeply, empathic human creatures capable of nuanced emotional intelligence.

As more brands experiment with bots and associate-free stores, and creating transaction-less interactions, the people who are on hand to help will need to be highly skilled and able to handle any situation or question that arises.

Some brands have already gone this route. Russia’s department store Au Pont Rouge restored their historic art-nouveau building and transformed parts of it into a sleek exhibition retail space. Store associates function more like curators with extensive product knowledge while a robotic system, powered by a mobile app, selects and transfers items to a payment area outside the exhibition space. The robot becomes the shop clerk, and the clerk becomes a subject matter and hospitality expert.

Or, take Apple’s concept of designing stores as “town squares.” By creating intentional community spaces, the store becomes a place to gather and create community, and it just so happens that you can shop there, too.

Create human-optimized interactions

Technology-driven changes in both physical and digital spaces may mean that we interact with humans less often, and some of those exchanges may feel uber-efficient—which is often the goal—but it doesn’t mean that we’re all going to be yelling at robots anytime soon. Instead, we have to think through the ways our technology can feel more human, and what we humans could gain from leaning on technology. Rather than replace us, it can help us to focus on what we do best: innovating, problem-solving, and building relationships. This last part—creating human connections—is exactly what leads to the kind of deeper, more personal customer relationships that keep customers coming back.

We have to think through the ways our technology can feel more human, and what we humans could gain from leaning on technology.

“You must adapt. Automation is coming and it’s a question of when, not if,” Ancketill said. Many new companies are basing their business models on AI, which will make it increasingly difficult for companies not using new technologies to compete. Those who leverage AI and machine learning now, however, will be ahead of the curve.

Ultimately, there’s a lot that we still don’t know, just as there’s a lot that Alexa still doesn’t know. For now. Yet we do know that the future is looking increasingly screenless and voice-activated and that products and experiences will become more and more tailored to the customer. Consumers will have a lot of choice, and they will invite some smart technologies into their lives, but not others. As technology gets even smarter, we’ll be pushed to grow and get smarter, too—about how we use technology to better serve our customers, and each other. In the meantime, Alexa powers a mean dance party.

Suzanne Barnecut loves reading and writing stories of all kinds and duration. She is a frequent contributor to Relate, and creates brand content and tells customer stories for Zendesk. In her spare time, she can be found writing fiction, reading The New Yorker, and consuming (too many) pastries from San Francisco’s bakeries. Find her on Twitter at: @elisesuz.