Join top executives in San Francisco on July 11-12, to hear how leaders are integrating and optimizing AI investments for success. Learn More

Last week Forrester published a pair of surveys on artificial intelligence — “Predictions 2019: Automation” and “Predictions 2019: Artificial Intelligence.” What stood out was a frank statistic that no business really wants to admit: In 2019, a full 10 percent of companies will bring human expertise back into the AI loop, and this trend will be due largely to realizations of AI’s limitations.

Let’s be clear: I’m talking about AI-driven automation for business here, not the advancements in AI which are, for example, saving lives. No one is arguing that AI’s mark on health care and training algorithms to detect cancer cells in images is worth taking steps back from. The doctor in the machine is best left to get on with it.

But when it comes to business, Forrester has uncovered some hard truths. No business wants to admit it’s getting AI wrong, and as a business leader of an AI-driven solutions organization myself, I’m curious about any company that experiences AI “fallout.” However, the AI retreat stat is hardly surprising, especially when you look at AI from a customer service point of view.

Time to curb AI enthusiasm?

What is surprising about Forrester’s reports is that only 10 percent of companies are apparently taking steps back from AI. Considering the vast number of companies that are automating the wrong things — and others that have completely unrealistic expectations of AI — I would expect to see 50 percent of respondents planning to take a step back in 2019 and bring some much needed humanity back into the loop.


Transform 2023

Join us in San Francisco on July 11-12, where top executives will share how they have integrated and optimized AI investments for success and avoided common pitfalls.


Register Now

Companies are beginning to see beyond AI’s economies of scale, realizing some of the hard truths the surveys have revealed; namely that automation is great, but realistically, it’s only great if it brings you closer to your customer. If AI is replacing the human on help desks or live chats, you’re onto a losing streak when it comes to customer service and satisfaction.

The truth, as well as the proof, is out there, and what it’s telling us is that most humans are keen on human contact. We’ve been interacting with one another for thousands of years now, and it turns out that customer service is one area where we expect human expertise in the loop.

The claim stands up to scrutiny. Look at the extensive research from Harvard Business School’s Ryan W. Buell on automation over the past 10 years. He’s proved that customers who use ATMs more than human tellers have lower levels of satisfaction with their banks, and his flagging of Metlife’s automated telephone condolences message to death-related insurance customer claims is surely a red flag that sympathy should not be relegated to AI.

Automate for the people

Perhaps the biggest lesson we can take away from the research is that, in most cases, humans shouldn’t be left out of the AI loop entirely. We need to automate for people in a way that shows we understand what our customers really want — and in a way that uses automation to boost customer support processes.

Striking the right balance between AI and humans usually means using automation as a first line of response. This appeals to the large majority of customers who will attempt (and even prefer) to take care of matters themselves before reaching out to a live representative anyway. As long as there is always an option to switch to a live representative, most businesses will find this an effective balance between automation and human contact.

A great example of this is automation that detects users having recurring difficulty in any step of the journey and triggers a chat window with a human customer support agent. Another example is having automation collect and deliver customer information to customer support agents before contact initiates so that actual conversations are informed and personalized.

It’s about using automation to help your customers secure the best, fastest, most personalized customer support delivered by a human when they need it.

2019 is fast approaching, and the research is well timed to remind us that AI is here to free us to have the good conversations with our customers – not the straightforward and easy stuff, but the real issues that ultimately build and drive brand loyalty. If we can keep humans in the loop and use AI to make customer support operations more efficient, we can focus on AI’s potential rather than its limitations.

Vasco Pedro is CEO of Unbabel.

VentureBeat's mission is to be a digital town square for technical decision-makers to gain knowledge about transformative enterprise technology and transact. Discover our Briefings.