If pop culture is to be believed, society is quickly heading toward a highly automated future ruled by artificial intelligence. Take Iron Man’s trusty sidekick, J.A.R.V.I.S. Within the Marvel franchise, the artificial intelligence system is able to think, act, and feel like a human. The supporting character is even sarcastic and witty — both trademark human characteristics. In some ways, J.A.R.V.I.S. seems like a better human than most humans.

With the release of AI technologies like IBM Watson and Salesforce Einstein, in addition to the recent buzz about the “Partnership on AI,” which has brought together some of the world’s biggest tech companies to advance research in the sector, it might seem like that fantasy is quickly turning into reality. It’s no wonder Forrester predicts that robots will replace 6 percent of U.S. jobs by 2021, with Salesforce’s Einstein promising to help close sales deals and Oracle’s Adaptive Intelligent Applications providing actionable business and customer insights across multiple industries.

While we’re seeing bold progress in the sciencesmedicine, and education, it’s unlikely we’ll all be governed by robots anytime soon. The truth is that fully autonomous AI is still science fiction, and these machines are still largely dependent on humans, especially when it comes to learning on their own and performing complex tasks.

Chatbots are useful, not powerful

Let’s consider the recent hype around chatbots. While we’ve seen enormous interest in all things bot lately, most chatbots today are generally limited to answering questions and handling simple transactions. And rather than acting as true self-learning machines, chatbots must be manually trained by humans in order to get smarter.

While it’s true that conversational technologies are evolving in ways that are reshaping how companies and consumers interact, discussions around the self-service revolution are more likely to be centered on striking the right balance between automation and human assistance, rather than predictions of total takeover like 2004’s I, Robot. The diversity of consumer intents and the complexity of business processes make fully automated machine learning impractical, at least for the near future, and humans remain entirely necessary.

Technological advancements have paved, and will continue to pave, the way for automated experiences, such as virtual assistants, to become the starting point for almost all interactions a customer might have with a business. In my lifetime, we’ve gone from hoping the newspaper printed the right movie times to asking Siri for movie schedules and purchasing tickets with the touch of a button. In the era of ubiquitous computing where computers outnumber users, the pop culture representation of intelligent computers seems possible.

But for now, despite the strides being made by powerhouse tech companies, we’re in the middle of an AI hype cycle with inflated expectations about how much these machines can accomplish without us.

Saving humans for high-value conversations

Gartner predicts that 85 percent of customer interactions will be managed without a human by 2020, so it’s clear that companies should and will continue to invest in bots. Still, the best consumer experience often requires a balance of human and machine, depending on the nature of the customer’s query and their ultimate intent. Flexibility and responsiveness are key for engaging customers, whether fulfilled by a human or a machine. The future is not about machines replacing humans entirely, but rather for saving human effort for those particularly unique or complex scenarios that require human understanding of natural language and a sense of empathy.

To date, we’ve yet to see any one company fully leverage available data to make customer engagement as effective as it can be. Consumer data comes from many disparate places and exists in many forms. The volume, speed, and diversity of data can be impossible for humans to comprehend. That’s where machines come in. From advanced natural language technology to prediction and machine learning, technology is critical to making sense of data and context in real-time in a way that helps businesses engage their customers in the moments that matter. And this ability will improve over time.

These technologies can help businesses determine the consumer’s intent, weigh their needs, and make more informed, personalized, and real-time service decisions. For example, a business could use AI to determine which channels a shopper prefers, whether they like to interact with a virtual assistant or a live chat representative, or if the customer’s request may require higher-level assistance. It’s the same way the waitstaff at a high-end restaurant might anticipate customers’ needs, appearing at just the right time or recommending the perfect wine.

Research shows that consumers’ first choice will almost always be to handle issues independently. In fact, one-third of millennials say the ability to self-serve is what they look for most often in a customer service experience.

While self-service technology is getting better at triangulating customer needs, many consumer engagement scenarios require human intervention. At times, the range of consumer preferences and emotions are simply too broad for even the smartest machine to address. In all cases, the switch from a bot to live agent should feel contextual and easy for the customer, and empower the live agent to successfully handle the customer’s request post-transition.

Self-service grocery lanes are a great example. Customers certainly enjoy the ease of self-service, and grocers benefit from the ability to process transactions faster. However, self-service grocery lanes cannot always scan a customer’s ID when buying alcohol, for example. At that point, cashiers need to step in to resolve the issue (without asking the customer to start over again!).

The same should be true of the customer experience online. In a world where the vast majority of customer interactions are automated, humans should be reserved for high-value, high-touch conversations that drive revenue and brand loyalty. It’s where machines fall short, and where humans are most valuable.

Hollywood is right, but ahead of its time

The future of self-service for companies is akin to the state of a self-driving car. We’ve made cars smarter and customer service more effective, but we’re still a long way from wide adoption of self-driving cars and chatbots to resolve every problem.

That’s why customer service of the future will continue to see humans and machine working together. J.A.R.V.I.S. is amazing, but as the acronym suggests, it’s just a very intelligent system. There are certain customer service needs that only humans can handle. While J.A.R.V.I.S. may inform Iron Man’s decisions and act incredibly human (much more so than today’s technologies), it can’t replace the man in the suit.